Bot configurations
Last updated
Last updated
The bot uses config.yaml file to load all the configurations.
Different configuration can greatly impact your profitability using the bot. Tweak it based on your own setup. Join our discord to learn more.
You can also refer the config examples as a starting point to tweak for your own config.
This section contains configurations like private keys, api keys.
Private key for your wallet. All formats are supported, you can export from Phantom/Solflare or just use a json file path. The wallet needs to hold enough SOL/WSOL/USDC for your configuration.
If present, JupV6 API will use it, otherwise JupV6 will be running in pull mode, which is very heavy on the read RPC.
If your yellowstone provider also provides you an x token set it here.
When running the bot with JUPITER_URL=http://0.0.0.0:18080, or JUPITER_URL=http://127.0.0.1:18080 the bot will first pull the latest popular token mints from Birdeye, based on last 24h volume. Then spin up a JupV6 API with FILTER_MARKETS_WITH_MINTS set to the token mints pulled from Birdeye. INTERMEDIUM_MINT_COUNT_FROM_BIRDEYE configures how many tokens to use from Birdeye. The higher you set for INTERMEDIUM_MINT_COUNT_FROM_BIRDEYE, the more oppotunity you may see as there are more potential routes from different token pairs. Though higher number also means longer time to get a quote from JupV6 API, and it will also uses more resources from your machine. If YELLOWSTONE_URL is not set, it is not recommended to set
If YELLOWSTONE_URL is set, it will be used to start JupV6 API with it, otherwise the JupV6 API will run in pull mode, which is heavy for your read RPC and it is not recommended when you have too many intermedium mints(>100).
If JUPITER_URLis set to a different value, none of the params in the following form will apply. This can be used when you have a JupV6 API provider, or running it yourself separately.
Default: false
When set to true, it will not start the Jupiter API locally. This is useful when you want to run multiple bots on the same machine
default: []
You can only set one of INCLUDED_DEXES and EXCLUDED_DEXES not both.
default: []
You should always exclude Moonshot and Pump.fun unless you know what you are doing.
You can only set one of INCLUDED_DEXES and EXCLUDED_DEXES not both.
The RPC that is used to read data. When using default JUPITER_URL value, this RPC will also be used to spin up JupV6 API. If this is not set, the value of RPC_URL will be used.
Default: http://0.0.0.0:18080
Endpoint for JupV6 API. When using default value, it will spin up a JupV6 using top popular mints from Birdeye. When set a value other than default, the rest of the parameters in this form will not be applied.
Default: 24
the maxAccounts param that is used to call quote from JupV6. Recommended not setting this higher than 32. The higher this number is, the more routes you can find, though as there's a limit on the total accounts, a higher number here will also cause a higher chance to find an invalid route with too many accounts, and it also takes longer to search.
Default: false
When this is set to true, it will load the market cache from the local file mainnet.json. This is useful when you run jup and it's very slow to start, or even fail with error reading a body from connection. You can update the local file with wget https://cache.jup.ag/markets?v=4 -O mainnet.json
Default: false
Default: 18080
Default: remote
Supported value: europa,file, remote
The default behavior(remote) here will fetch a snapshot of all the markets given your intermedium mints, and will only update them on a restart. Europa mode will stream new markets for your mint list to Jupiter, but it can be very heavy for some RPC to handle.
Default: the number of CPU cores on your machine
Routing engine thread = TOTAL_THREAD_COUNT - WEBSERVER_THREAD_COUNT - UPDATE_THREAD_COUNT.
Most of the time you shouldn't need to set this and can just let jupiter use all your cores to enhance performance.
Default: 2
This is how many thread to handle the http requests. The bot is heavily using this so you should consider setting this to a high number if your jupiter quote per second is high.
Default: 4
This is the number of threads used to handle markets update using the data pulled from yellowstone/rpc. If you are using a lot of mints you should consider setting a higher number for this to speed up your price update.
Default: true
When set to true, the bot will only look for 2hop(A -> B -> A) trade. Setting to true will lead to shorter quote time to Jup, with faster landing. Setting to false will lead to more opportunity, but taking more time to quote.
You should only set this when using jupiter from a third party provider, and are using x-api-key in the header(like shyft), you can set it here.
If your provider uses a url params, like api-key=xxxx in the url, please use ADDITIONAL_JUPITER_URL_PARAMS
If your provider uses a url params, like api-key=xxxx in the url, you can set this config to include that in all the jupiter requests.
example:
Some providers does not expose the /tokens api endpoint for their jupiter service. In such case you can set this config to true to avoid the call. Otherwise the bot will stuck at Fetching Jupiter tokens from
Default: false
From Jupiter: Restrict intermediate tokens to a top token set that has stable liquidity. This will help to ease potential high slippage error rate when swapping with minimal impact on pricing.
The Bot tries to find profitable route by getting a quote from BASE_MINT to an INTERMEDIUM_MINT for a given trade size, then get another quote from INTERMEDIUM_MINT to BASE_MINT, and combine them together for a potential profitable route. Basically a route from A -> B -> A.
When USE_DIRECT_ROUTE_ONLY is set to true, the bot will only return route in the form of A -> B -> A like above. For potentially more opportunity, you can set it to false, which can lead to routes like A -> B -> C -> A. The downside of setting this to false is it will slow down the time to get a quote, end up sending out the transaction a bit slower.
You can set as many tokens as possible as the INTERMEDIUM_MINT, though the more you have, the heavier your RPC load is, so you may want to adjust this accordingly.
There are multiple ways for you to define which mints you use as intermedium. The bot will merge all the mints fetched through different ways into one list and use it as intermedium mints.
Default: 0
This param configs how many top tokens to use as intermedium mint that pulled from Birdeye's 24h volume list. This list has high volume but usually low velocity.
Default: 0
This param configs how many top tokens to use as intermedium mint that pulled from Birdeye's trending list. Maximum 200. Update every 20min. This list has high velocity new mints, and usually works better with SOL as base mint.
You do not need birdeye api for this.
Default: 0
This param configs how many top tokens to use as intermedium mint that pulled from Jupiter's trending list.
Default: 0
This provides a way to load intermedium mints for the bots to use. This is useful if you have a set of mints in mind that you want to use, or you write your own script to fetch intermedium tokens for the bot to use.
Example file:
Same as INTERMEDIUM_MINT_FROM_FILE except this take in a list of files.
Example:
Just a list of mints. This is convenient when you just want to manually specify a few mints to run, you can do this without creating another file.
Example:
This provides a way to exclude certain mints from intermedium mints.
This is useful when you want to exclude certain mints from birdeye/gecko. Especially if you only want to do SOL arbs and don't want to include USDC/USDT in the route, as it can be heavy for your RPC to load all.
The following example is to exclude USDC/USDT from your intermedium mints.
This allows you to copy trade top players and use whatever intermedium mints they use.
For example, if you want to copy ben/bob's most recent 100 tx:
You can define an overall MAX_MINT_COUNT, and a MAX_MINT_COUNT for every address you are tracking. The bot will rank all the mints by number of times it appears and keep at most MAX_MINT_COUNT mints. If you have INTERMEDIUM_MINT_EXCLUDE_FROM_FILE defined, the mints will be excluded first then cut off the top rank.
This is a useful feature when you run multiple machines and want them to run different mints. You can use all the other ways to fetch, and use this filter to randomly choose a few for all your machines.
This section contains configurations for the bot itself.
Default: 60
Unit in minutes. By default, the bot will auto restart every 60min. This will also restart JupV6 if hosted together, to allow fetching the lasted high volume tokens from birdeye. If set to 0, the bot will not auto restart.
Default: true
Whether to run simulateTransaction before sending the transaction. This can help to reduce the number of failed transactions landed on chain if spam mode is enabled.
Default: false
DO NOT SET THIS TO TRUE if you don't know what is a Jito RPC. You have to use a Jito RPC when you enable this. When this is enabled, the bot will use simulateBundle to run simulation, which can give you the actual profit with the simulation. This value is more accurate than then one calculated from Jupiter quote. This is especially useful when you set spamFeeBps in your base-mint.json
Note: Most providers don't provide Jito RPC. Right now the only options to get one is to either host a RPC node yourself, or rent a dedicated RPC node from a provider like Helius/Triton.
Default: true
By default the bot will encrypt your private key on first run, and ask you to remove your private key for better security practice. You can turn off this feature but it's highly not recommended.
Default: 0
Default: 1
Bot will use an estimated value for compute unit limit based on the route. The default value can make sure 95%+ of the tx have enough compute unit.
But this can be too big/too small based on different JUP_MAX_ACCOUNT, you can adjust this value to higher/lower as needed.
Default: 400
The maximum compute time to send tx for an opportunity. If you are trying to catch an opportunity right after it appears, you usually don't want to send anything when it takes too long to calculate in the first place. If the time took to compute is greater than MAX_COMPUTE_TIME, the bot won't send any tx for it.
Default: 20_000_000
The max compute unit price you set to wrap/unwrap SOL. This can be useful when the network is super congested.
Text that will be added as the last instruction in the transaction. You can use this to distinguish which server you are running the bot from by setting a different value for each server.
When not set a MEMO will NOT be added to your tx.
Default: false
Default: 0
Units in lamports. Does nothing by default. When set to a none 0 value, the bot will periodically check the wallet's SOL balance, and when it drops below the value, it will unwrap all the WSOL and wrap only the amount needed back, to rebalance the SOL account to avoid running out of SOL. e.g. set this to 1000000000 will auto balance your wsol/sol accounts when your sol drops below 1.
The RPC that is used to read data. When using default JUPITER_URL value, this RPC will also be used to spin up JupV6 API.
Default: false
When enabled and you are using fixed trade size(either using tradeSize or tradeRange with linear strategy), the bot will check if it's the same as last time and not generate a new opportunity for it if it's the same.
Default: false
This will only affect SOL as base mint. When enabled, the bot will unwrap your profit before sending the jito tip/bot fee, which can be very useful if you don't leave much SOL balance in your wallet, or if you are targeting larget opportunities and afraid to not have enough SOL to pay for a big hit.
Default: 1
Config how many worker(threads) to use when fetching quotes from Jupiter.
Default: ''
Default: 0
When set to a none 0 value, it will apply a rate limit when doing RPC calls. This is helpful when using the free plan for some RPC.
Default: false
When you use blind quote mode, the bot will quote the first half of the transaction from SOL->X, then quote the second half X->SOL. The second quote need to wait for the first one's result before it can start. For fixed size, you can enable this optimization, which will use the quote result from the previous round's quote as an estimation, so the SOL->X and X->SOL quote can be sent out simultaneously. This will result in a shorter Jupiter quote time. But occasionally the two quote will find the same pool, which makes the oppotunity invalid, this is especially true when you are trading a new mint where it only has limited pools.
Following is the Jito related configs you can set in your config.yaml file
Default: true
Whether to use Jito bundle to send transaction. The good part about Jito is, it will never land if the transaction fails. So everything lands is profit.
Default: false
Whether to use jito GRPC to send bundles. This is NOT recommended unless you have a key with high rate limit.
NOTE: BLOCK_ENGINES does not apply to jito GRPC. You need to set BLOCK_ENGINE_URL instead, and you can only send to one region with GRPC.
Only apply to Jito GRPC.
The block engine url you want to use with your jito GRPC.
e.g. amsterdam.mainnet.block-engine.jito.wtf
Only apply to Jito GRPC.
The key json file you want to use with your jito GRPC.
e.g. jito.json
Default: ny,amsterdam,frankfurt,tokyo,slc
You should not change this value unless you really know what you are doing. Config which block engines to send to for jito transactions.
Default value: ['dynamic']
Supported values: dynamic, static, static-list, fixed, fixed-list, fixed-range
Define which strategy to use when paying for jito's tip. DO NOT use static and fixed together before version 1.3.16.
static-list: Similar to static, but you send multiple tx with different static bp for the same opportunity. You need to set STATIC_TIP_BP_LIST for this to work.
fixed-range: Similar to fixed-list, but you instead of you provide the tip amount, bot will automatically generate it for you. You need to define FIXED_TIP_RANGE, FIXED_TIP_RANGE_COUNT, FIXED_TIP_RANGE_STRATEGY for this to work.
Default: 0
Jito tip as part of your profit. 5000 means 50% of your profit will be used as Jito tip. The higher this number, the better chance you may land your transaction, but in return, less profit earned.
Default: 0
Jito tip as part of your profit. 5000 means 50% of your projected profit will be used as Jito tip. The higher this number, the better chance you may land your transaction, but in return, less profit earned.
The tx landed with this may not have the same percentage as you defined, as the tip value is set before you send the tx, but the actual profit may vary when it actually landed.
Sending multiple tx with different bp.
Example
Default: 0
Use a static jito tip when sending to Jito. Unit is SOL lamports. The bot will always use the same tip for all tx it sends, and still make sure the tx landed are all profitable.
When you set both DYNAMIC_TIP_BP and FIXED_TIP, bot will send both percentage based and fixed tip transactions.
Note: If you are using this without USE_SEPARATE_TIP_ACCOUNT set to true, the bot will use sendTransaction endpoint to sent it, which is using a different rate limit then sendBundle which is used with DYNAMIC_TIP_BP.
Default: []
Example:
Use a static jito tip when sending to Jito. Unit is SOL lamports. The bot will always use the same tip for all tx it sends, and still make sure the tx landed are all profitable.
When you set both DYNAMIC_TIP_BP and FIXED_TIP_LIST, bot will send both percentage based and fixed tip transactions.
Note: If you are using this without USE_SEPARATE_TIP_ACCOUNT set to true, the bot will use sendTransaction endpoint to sent it, which is using a different rate limit then sendBundle which is used with DYNAMIC_TIP_BP.
Define a range for your tip. Bot will generate a list of different tips based on the strategy you choose.
Example:
How many different tip you want to generate for your tip range.
Valid values: random, exponential, linear
linear will generate trade sizes evenly within the range.
random will generate a trade size randomly each time.
exponential will generate trade size on exponential values. For example for range 10 - 1000, with count 5, it will generate trade size like this: [10, 77, 315, 668, 1000]
Default: 0
Set the max tip the bot pays when using dynamic or static jito tip trategy.
Unit is SOL lamports. The bot also has a max tip amount of 80% of your remaining SOL balance, to avoid missing big opportunities when not having enough SOL left. 0 means no no max tip.
By default, jito has a 5/s/ip rate limit. You can ask for a rate limit raise from jito, and they will give you an uuid that you can set with the bot.
NOTE: Currently the key has a bug and the rate limit for the key is universal not per ip.
Default: false
When this is enabled, response for sending jito bundles will be added to the output log. This is helpful if you are trying to figure out how often you are hitting jito's rate limit, or testing your UUID to see if it works.
Default: false
Default: false
This will send different TX for different block engines, in theory increase your chance to land, and can some time land multiple tx for the same opportunity.
Default: false
This will use different tip account for different block engines, in theory increase your chance to land, and can some time land multiple tx for the same opportunity.
Default: false
This is the build in support for multiple ips on your machine. When this is enabled, bot will scan all your ip and round robin it when sending through jito. You can also define an ip range with START_IP and END_IP. Or IP_LIST to manually define the list of ips to use.
Only applied when you have ENABLE_MULTIPLE_IP_ADDRESSES_FOR_JITO set to true.
The start ip for bot to use, in IPV4 format.
E.G.
Only applied when you have ENABLE_MULTIPLE_IP_ADDRESSES_FOR_JITO set to true.
The end ip for bot to use, in IPV4 format.
E.G.
Define the exact ips you want to use for ENABLE_MULTIPLE_IP_ADDRESSES_FOR_JITO .
Example:
Spam strategy means you will send many transactions and hope some of them get landed and landed successful. This is a more advanced strategy as you may lose money on gas fee if you didn't config it carefully.
You can enable spam by setting ENABLE_SPAM=true in your config.yaml file. The bot will then use all the rpcs you defined for SEND_RPC_URLS to send potential profitable transactions.
Jito and spam can be enabled together, the bot will just send 2 different transactions, one through jito and one through rpc.
Default: false
Whether to send a normal transaction for every potential arbitrage oppotunity. Note most of the transactions will not even land on chain, for the ones that does, 90% of them will fail. Set ENABLE_SIMULATION=true can help reduce the number of failed transactions.
Default: 0
Default: 0
Default: fixed
Supported strategies: profit-based, fixed, helius. You can only use one of them.
fixed: This will use the value of COMPUTE_UNIT_PRICE setting to set the compute unit price value.
Default: 600
The RPCs used to send transactions. This can be the same as RPC_URL
You can also set multiple RPCs here, split by comma. Setting multiple ones can increase the rate of tx landing and success.
Example:
When this is set, it will override the COMPUTE_UNIT_PRICE value in config.yaml file.
This should be set together with MAX_COMPUTE_UNIT_PRICE. The max COMPUTE_UNIT_PRICE you are willing to set. This is helpful to prevent setting a huge priority fee as market moves.
Example usage of the config:
This allows you to set your gas fee to be a fixed percentage of the projected gain. Note you can only set one of PRIORITY_FEE_PERCENTILE and SPAM_FEE_BPS, not both.
Example usage of the config:
Default: 0
For helius and profit-based PRIORITY_FEE_STRATEGY, this field is required. This is to prevent you from paying some absurd amount of gas fee.
Default: 0
For helius and profit-based PRIORITY_FEE_STRATEGY, you can set a min compute unit price.
Default: true
This is a similar config to ENABLE_SIMULATION. The difference is ENABLE_SIMULATION will run the simulation with your RPC_URL before sending it, when you have SKIP_PREFLIGHT=false, the RPC received your transaction will run the simulation before it forward to the validator. This will also make landing slower as it's an additional call, but also it can reduce the failed transactions you land with ENABLE_SPAM.
You should only have at most one of SKIP_PREFLIGHT=false and ENABLE_SIMULATION=true. Setting both is not worth it.
Default: 0
Valid values are 0,1,2,3,4,5.
This config will try to limit how long a tx land compares to when you send it when using spam. 0 is no restriction, the same behavior as before. 5 is maximum restriction and will only allow the tx to land within ~1-2 slots. Note the exact restriction time for each level can be affected by your RPC's latency and performance. You may not land anything when you set this value to 4-5 if your RPC is not fast enough.
This config allows you to use a very high fee for spam strategy and also not losing too much money on failed ones.
Default: false
Default: false
Note when a tx landed successfully with BloXroute, it charges an additional 0.0005 SOL as fee to use BloXroute's api and send through its staked connection. The bot already takes this into consideration and will fail the tx if your profit is not enough to pay for this fee + the bot fee. But do take this into consideration when you enable it.
You need to have your own key to use BloXroute to send.
Default: ny
Only relevant when you have ENABLE_BLOXROUTE=true. Valid values are ny/uk/la/tokyo/germany/amsterdam. Should only set it to be the closest one to your server.
If you want to send to multiple region for the same tx, you can use this.
example:
Default: ''
This is required to use Bloxroute.
Default: 500_000
The bloxroute fee you pay, in theory the high this is, the faster you tx can be delivered. Minimum is 500_000.
Default: false
Use dynamic fee for Bloxroute instead the fixed value from BLOXROUTE_FEE. The fee will calculate the fee based on the projected profit similar to STATIC_TIP_BP in Jito. You can set the base points used with BLOXROUTE_DYNAMIC_FEE_BP
If the calculated fee is smaller than BLOXROUTE_FEE, BLOXROUTE_FEE will be used as the fee.
Default: 0
Calculate BloXRoute fee based on your projected profit. 500 means 5% of your projected profit will be used as the fee paid to BloXRoute.
The tx landed with this may not have the same percentage as you defined, as the tip value is set before you send the tx, but the actual profit may vary when it actually landed.
Default: 0
Maximum dynamic fee in lamports, 0 to disable
Default: false
When enabled, bloxroute will only send through jito. aka you won't see failed tx with bloxroute.
There's no default key for NextBlock, you need to set your own key to use this.
Default: false
Default: ny
Supported region: ny or fra
If you want to send to multiple region for the same tx, you can use this.
example:
You need to set this to use NextBlock
Default: 1000000
The fee you pay to nextblock for each successful transaction
Default: false
Use dynamic fee for NextBlock instead the fixed value from NEXTBLOCK_FEE. The fee will calculate the fee based on the projected profit similar to STATIC_TIP_BP in Jito. You can set the base points used with NEXTBLOCK_DYNAMIC_FEE_BP
If the calculated fee is smaller than NEXTBLOCK_FEE, NEXTBLOCK_FEE will be used as the fee.
Default: 0
Calculate NextBlock fee based on your projected profit. 500 means 5% of your projected profit will be used as the fee paid to NextBlock.
The tx landed with this may not have the same percentage as you defined, as the tip value is set before you send the tx, but the actual profit may vary when it actually landed.
Default: 0
Maximum dynamic fee in lamports, 0 to disable
Default: true
Show NextBlock sending error messages in logs
Default: false
When enabled, Nextblock will only send through jito. aka you won't see failed tx with Nextblock.
There's no default key for Temporal, you need to set your own key to use this.
Default: false
Default: pit1
Supported region: pit1 ,ams1, fra2 ,ewr1
If you want to send to multiple region for the same tx, you can use this.
example:
You need to set this to use Temporal. Get one from their website.
Default: 1_000_000
The fee you pay to temporal for each successful transaction
Default: false
Use dynamic fee for Temporal instead the fixed value from TEMPORAL_FEE. The fee will calculate the fee based on the projected profit similar to STATIC_TIP_BP in Jito. You can set the base points used with TEMPORAL_DYNAMIC_FEE_BP
If the calculated fee is smaller than TEMPORAL_FEE, TEMPORAL_FEE will be used as the fee.
Default: 0
Calculate Temporal fee based on your projected profit. 500 means 5% of your projected profit will be used as the fee paid to Temporal.
The tx landed with this may not have the same percentage as you defined, as the tip value is set before you send the tx, but the actual profit may vary when it actually landed.
Default: 0
Maximum dynamic fee in lamports, 0 to disable
Default: true
Show temporal sending error messages in logs
Fast currently only has one region amsterdam.
Default: false
Default: 1_000_000
The fee you pay to fast for each successful transaction
Default: false
Use dynamic fee for Fast instead the fixed value from FAST_FEE. The fee will calculate the fee based on the projected profit similar to STATIC_TIP_BP in Jito. You can set the base points used with FAST_DYNAMIC_FEE_BP
If the calculated fee is smaller than FAST_FEE, FAST_FEE will be used as the fee.
Default: 0
Calculate Fast fee based on your projected profit. 500 means 5% of your projected profit will be used as the fee paid to Fast.
The tx landed with this may not have the same percentage as you defined, as the tip value is set before you send the tx, but the actual profit may vary when it actually landed.
Default: 0
Maximum dynamic fee in lamports, 0 to disable
Default: true
Show fast sending error messages in logs
Default: false
When enabled, Fast will only send through jito. aka you won't see failed tx with Fast.
Default: false
Enable debug mode for additional logging
Default: false
When this is enabled, it will output an additional WIN log when a transaction is landed. Though if you turn this on while having ENABLE_SIMULATION=false, it may consume a lot of your RPC credits.
Default: true
When this is enabled, it will output all the tx the bot sent out, through which provider(jito/rpc/bloxroute etc.). You can turn this off if the log is too overwhelming to you.
Default: false
If you don't want to see the error messages, you can turn this to true which will hide them.
Flashloan is a power tool when you have limited budget. You can set a higher trade size than what you have in your wallet, and get arbitrage opportunities that you usually won't be able to get.
Flashloan may not success 100% of the time. When the vault you are borrowing from does not have enough tokens, your transaction will fail due to insufficient fund. Though this should rarely happen.
When flashloan is enabled with the bot, and you are using SOL as one of your base mint, you should also set WRAP_SOL_AMOUNT which is the amount of sol you want to wrap into WSOL. The bot will only use flashloan when the amount is larger than WRAP_SOL_AMOUNT, for trade with less amount it will just use the amount in the wallet, to maximize the potential profit.
Default: false
Whether to enable flashloan. Note you need to have at least 0.5 SOL in the wallet to enable flashloan.
Default: kamino
Currently supported values are solend, kamino
Kamino is the recommended provider. It does not charge any fee and the liquidity pool is very deep so you can easily borrow up to a few million.
Soland flashloan charges a 0.04% fee based on the amount you borrow. So a trade that borrows 1000 USDC as shown above will need to find an opportunity that gives you at least 1000.4 USDC back for you to make a profit, compares to at least 1000 USDC when you are not using flashloan. This will results in more failed transactions if you send using spam, or less overall landing if use with Jito. The upside is you can get opportunities that is only possible with big amount. You can also use flash loan as a kick start option to minimize your risk of putting too much amount in your wallet.
Solend only adds 2 more instructions instead of 4. So you will run into less Generated transaction is too large error compares to marginfi, but paying an additional fee as a trade off.
Default: 0
The amount of SOL to wrap into WSOL, only relevant when SOL is used as one of the base mint. By default the bot will wrap the largest amount in your trade size, but when flashloan is enabled you have to define how much you wrap. For all the trade with trade size less than this number, the bot will use the balance in the wallet, for larger ones the bot will use flashloan.
Blind quote strategy is one way to find opportunities with the bot. You define a list of intermedium mints, the base mints you want to trade with(SOL/USDC/USDT), and what trade size you want to trade with. The bot will use your trade size to quote for all the intermedium mints, one buy and one sell, and find profitable routes by combining the buy and sell routes together.
This strategy is a good start for beginner. It's easy to land as it can combine a lot of small opportunities into a big one, and generally less competitive.
You have to specifically set this to true to enable the strategy
This is used to control what base mint you want to use for your trades and what trade size you want to use. Base mint means the token you are going to trade and profit with, currently only SOL/USDC/USDT is supported.
The following setup means the bot will run with SOL and USDC. Note the unit is lamports here. So 1 SOL = 1,000,000,000 lamports. 1 USDC = 1,000,000 lamports. 1 USDT = 1,000,000 lamports.
Using 1 SOL/2.5 SOL/5 SOL as trade size for SOL, and will only try to send the transaction when the opportunity provides at least 0.0005 SOL in profit.
You can use the same mints for different TRADE_SIZES/MIN_QUOTE_PROFIT pair to customize your strategy.
Example:
Default: 1000
Unit in ms. The default value means the bot will wait for 1000ms then try again for the same token-trade size pair. Reducing this number will increase the chance to catch price change right away, but increase the read RPC load.
MIN_QUOTE_PROFIT is the threshold for when to send a transaction for the potential profit, it does not guarantee that the actual profit land on chain will be that. The actual profit landed on chain can vary, and the bot will fail all transactions that ends up with a negative profit, and land all the ones with a profit, even if the profit is smaller than MIN_QUOTE_PROFIT.
If one opportunity is found and tx is sent out, wait at least this time(ms) before sending another one.
Set in lamports on what sizes you want to trade with. Bot will quote jupiter api with the exact trade size you defined here periodically trying to find the most profitable routes.
Example:
If you don't want to manually define every trade size, you can utilize the trade range parameters. To utilize this you need to define 3 params: TRADE_RANGE, TRADE_RANGE_COUNT, TRADE_RANGE_STRATEGY
TRADE_RANGE: This defines the lower bound and higher bound for the range(inclusive). Bot will generate trade sizes only within this range. So make sure you have enough balance of the token for the upper bound.
TRADE_RANGE_COUNT: This defines how many different trade size will be generated within the range.
TRADE_RANGE_STRATEGY: Three values are supported right now: linear , random ,exponential.
linear will generate trade sizes evenly within the range.
random will generate a trade size randomly each time.
exponential will generate trade size on exponential values. For example for range 10 - 1000, with count 5, it will generate trade size like this: [10, 77, 315, 668, 1000]
Note: quote using random is slightly slower than quote using linear or fixed trade size. As for random two quotes are sent out one by one while for fixed size two quotes are sent out together.
Default: 2hop
Valid values: 2hop, 3hop
By default, only 2hop routes are calculated.
A list of mints you want to use for your 3hop routes. The most common use case for this is to set USDC when you use SOL as base mint, or set SOL when you use USDC as base mint.
You can override JITO/SPAM configs for each base mint you define, this allows more complex strategy setups without the need to run a different instance of the bot.
Here's an example of overriding DYNAMIC_TIP_BP for two base mints. The one without any override will use the global defined value, the one with override will use the override value.
Note: Only JITO/SPAM configs support config override.
Backrun strategy is monitoring large trade onchain, and counter trade the ones found to find arbitrage opportunities. This is a more advanced strategy, and requires yellowstone to run.
Backrun strategy works like the following:
1. Bot will monitor the markets you defined in DEXES_TO_MONITOR, with all the intermedium mints you defined above
2. If a trade cause the pool's price moved more than MIN_PRICE_IMPACT_PCT, bot will try to create a trade to backrun it
3. Bot will use the trade size you defined in TRADE_CONFIGS, to randomly generate a size between MIN_TRADE_BP and MAX_TRADE_BP, based on the trade you are back running. Or you can just define a FIXED_SIZE as your trade size here.
4. Bot will quote the other size of the trade with jupiter, create a full arb trade tx
5. If the tx has profit and it's higher than MIN_QUOTE_PROFIT, bot will send it out
Example config:
You have to specifically set this to true to enable the strategy
You can add memo to all the tx generated from backrun strategy. This will override the BOT level memo.
This will append the slot of the tx you backrun to your memo, you can compare this value to the slot number you tx landed on, to get a sense of how fast you land your tx.
Which dexes to monitor for your intermedium mints. Supported values are:
This will filter out trades that only has a price impact over this number. This is useful as most of the small trades won't generate an arbitrage opportunity.
This is used to control what base mint you want to use for your trades. Base mint means the token you are going to trade and profit with, currently only SOL/USDC is supported.
MIN_QUOTE_PROFIT is the threshold for when to send a transaction for the potential profit, it does not guarantee that the actual profit land on chain will be that. The actual profit landed on chain can vary, and the bot will fail all transactions that ends up with a negative profit, and land all the ones with a profit, even if the profit is smaller than MIN_QUOTE_PROFIT.
Unit in lamports. If the generated opportunity has a trade size smaller than this value, it will be skipped.
Unit in lamports. If the generated opportunity has a trade size bigger than this value, it will be capped at this value instead.
This is a list that contains two value MIN_TRADE_BP and MAX_TRADE_BP.
Let's say you see a trade trading 100 SOL for WIF. And you want to counter trade it. Usually you don't want to trade back the full amount as it may not have the best price for you. This config allows you to set a range, the example below means the bot will randomly generate a trade size that is 20% - 30% of the original trade. In this case the generated backrun trade will has a trade size between 20 SOL and 30 SOL.
Note this is a list, so you can define multiple range so that you can capture one trade with multiple tx with different trade sizes.
A list of route types you want to run the bot with. Currently the only supported type is 2hop, which means a trade like A->B->A. More types will be added later.
This is used for fetching lated priority fee. You can get one for free by register an account on .
This can be used to fetch tokens, fetch custom market cache etc.
is optional, but having it can greatly reduce your RPC calls and increase the number of mints you can have with Jupiter.
More info about self-hosted JupV6 can be found here:
The full list is . You can choose which Dex to include from your jup api when starting it. Using the address of the Dex not the name, separate by comma.
The full list is . You can choose which Dex to exclude from your jup api when starting it. Using the address of the Dex not the name, separate by comma.
You need to set to use this.
You also need to set to file and to true when using circular market cache
The port jupiter api uses. Note if you change this, you should also change your with the same port.
This only takes effect when you have to false.
This can help to speed up the jupiter quote time when you has set to false. Though as it says it will also restrict what additional mints it uses in your route.
Fetch new pools opened through gecko terminal and extract the mints used in the pool. API used here: .
Note can only return ~20-30 mints at maximum due to their limitation on the api.
If you have API key from , you can directly set it with the bot. It's the exact param as their document.
You can use a fixed COMPUTE_UNIT_LIMIT to avoid the additional computation needed. When this is set to 0, the bot will call jup to get an estimation of the COMPUTE_UNIT_LIMIT and use that instead. Which is more accurate but a bit slower, and will make 2 additional rpc calls getRecentPrioritizationFees and simulateTransaction for each quote.
A suggested value is 250000 for USE_DIRECT_ROUTE_ONLY=true and 350000 for USE_DIRECT_ROUTE_ONLY=false
The suggested value should work for 90% of the cases.
Note: when using flashloan, the bot will add 100000 to the value to account for the additional instructions needed.
You should not set this to be less than 200000, it may result in not landing anything successfully.
Read more about what is compute unit limit .
when enabled, will use your Yellowstone to pull blockhash instead of RPC. This can save you a lot of RPC credits. You must also set to use this.
When set, ths RPC url will be used when loading . This is useful when your main RPC does not support transaction history.
Tested you can use a helius free RPC plan with set to 5 to download all the intermedium mint from addresses.
Jito provides a way to send transactions through . Bundles will only land when it succeeds, which is perfect for making arbitrage trade!
You can enable jito by setting ENABLE_JITO=true in your config.yaml file. The bot will then send all transactions to all jito regions defined in through Jito's http endpoint.
This now also accepts urls, if you have a custom jito endpoint you can set it here. For example if you want to use url for the official ny endpoint, you can use
dynamic: The onchain program will calculate your real profit and use a percentage defined with to pay for the jito tip. This always send through the endpoint of jito.
static: This will use a percentage defined with based on the projected profit(the one you see in your opportunity) to pay for the jito tip. This means the actual percentage of the profit when the tx land may not be the same as the projected profit. This will use the endpoint by default, and when is enabled.
fixed: This will use a fixed number(SOL lamports) as your jito profit with . This will use the endpoint by default, and when is enabled.
fixed-list: Similar to fixed, but you send multiple tx with different fixed tip for the same opportunity. You need to set for this to work.
Note: and does not share the same rate limit. they both have 1/ip/wallet/region/s rate limit.
Your need to has dynamic in it for this to take effect.
Your need to has static in it for this to take effect.
Your need to has static-list in it for this to take effect.
Your need to has fixed in it for this to take effect.
Example with a FIXED_TIP of 20000 :
Your need to has fixed-list in it for this to take effect.
Your need to has fixed-range in it for this to take effect.
When this is enabled, a new solana address will be used to send tip in its own tx instead of sending to Jito directly for every Jito transaction. Jito currently is rate limiting bundles based on the accounts in the tx that does the tip, this feature can get around this rate limit by wallet issue to allow you using the same wallet in all locations. This is useful when you have machines in multiple locations but still trying to use the same wallet to run the bot. Example transaction:
The COMPUTE_UNIT_PRICE aka gas fee set for the transaction. Note: this is not SOL lamport. Generally the higher this is, the better chance your tx can actually land on chain. Learn more about what this is here: This only applies to spam, jito transactions set its own value as it's not needed.
This is a config for RPC method, Maximum number of times for the RPC node to retry sending the transaction to the leader. The higher this number, the more chance your tx can land, but also lead to high chance of failure, as retried tx is likely too late for the opportunity.
helius: When using this, you also need to define , and . This will pull the fee dynamically from helius and use a certain percentile value from it.
profit-based: You can only use this when you have and . This will use the value of SPAM_FEE_BPS to decide what fee to use for your tx. 100 SPAM_FEE_BPS means you will pay 1% of your simulated profit as the fee for this tx. This is a very high risk strategy, and can burn you a lot of SOL very quickly. But can also be rewarding when you hit big like this .
Unit in seconds. This controls how often the priority fee is updated. This only works when you set in your base-mint.json
This is an example instruction when setting =true
This image shows that the second RPC in the SEND_RPC_URLS list sent this tx(first RPC is RPC 0).
Value from 0 to 100(you can set higher than 100 but not recommended). Automatically set COMPUTE_UNIT_PRICE based on current priority fee level. You need to set HELIUS_API_KEY in order to use this. The bot utilize helius's , to fetch the latest priority fee levels. The response is something like the following from the api, value map to the response, aka 0 means min, 25 means low, 50 means medium, 75 means high, 100 means veryHigh. If the current response looks like below, and you set 30 as the value for priorityFeePercentile, the auto generated COMPUTE_UNIT_PRICE would be 10000 + (60082 - 10000) * ((30 - 25) / 25) = 20016. Note the value for each level will change all the time as market moves.
This config will use 5% of projected profit as the fee. Note this is not 5% of the final profit, it's 5% of the projected profit that bot calculated based on Jup API. If you set , this will be based on the simulated profit, which is more accurate than the projected one.
When setting multiple , this config will send a different tx for each of the RPCs. When this config is set to false, the same tx will be send to all RPCs. You can also see which RPC send the tx in the last instruction of the tx. Index start from 0. See later section for example*.
is a service provider that provides service to send transactions through their endpoint. Our bot has a special collaboration with them to provide you this service at a discounted rate.
You need to set to true when enabling this. Make sure you set the right value for when enabling this.
We have a collaboration with BloXroute, when this is set to true, your spam transaction will be sent through , in addition to all your SEND_RPC_URLS.
is a service provider that provides service to send transactions through their endpoint.
You need to set to true when enabling this. Make sure you set the right value for when enabling this.
is a service provider that provides service to send transactions through their endpoint.
You need to set to true when enabling this. Make sure you set the right value for when enabling this.
is a service provider that provides service to send transactions through their endpoint.
There's no default key for Fast, you need to set your own key and set it with
You need to set to true when enabling this.
Example:
Example:
This config only works with you have both and . Instead of using the quote profit that generated by jup, which can be very inaccurate sometimes, this allows you to use the simulated profit which is the exactly amount you will get if you land it right at your simulation time.
When this is set to true, no profit check for quote will be used. If you have simulation on the bot will still use that and only keep the ones that passed simulation. Jupiter quote can be very inaccurate sometimes so combining this with with ENABLE_SIMULATION can find you more opportunities. This can also be used as part of .
2hop: This means only routes like A -> B -> A will be considered. If you also set to false, there may be other mints between A->B and B->A.
3hop: This means only route like A -> B -> C -> A will be considered. You need to set which additional mints you want to include in the middle in addition to all the intermedium mints. The most common use case for this is to quote for route with SOL and USDC, like SOL -> USDC -> X -> SOL. 3hop will generate similar results like setting to false, but with much faster quote time.
You need to also define when using 3hop.
This config only works with you have both and . Instead of using the quote profit that generated by jup, which can be very inaccurate sometimes, this allows you to use the simulated profit which is the exactly amount you will get if you land it right at your simulation time.
When this is set to true, no profit check for quote will be used. If you have simulation on the bot will still use that and only keep the ones that passed simulation. Jupiter quote can be very inaccurate sometimes so combining this with with ENABLE_SIMULATION can find you more opportunities. This can also be used as part of .
