Comment on page

Loading and streaming events into a pool

Suppose a quant developer is going to backtest some strategy on mainnet pool.
This includes the following steps:
  1. 1.
    Using mainnet data to initialize a core pool
  2. 2.
    Replaying events up to a certain block as a checkpoint
  3. 3.
    Do some interaction as the strategy asks as interpolation to real transactions
  4. 4.
    Replaying events until the next checkpoint
  5. 5.
    Repeat step3-4 until events are run out
  6. 6.
    Evaluate performance of the strategy
Tuner has offered an example project called uniswap-v3-bot. It similarly follow the process above to build a strategy platform for backtesting, dry-run and run, where a user adds a new strategy by implementing some callback interfaces(trigger, cache, act and evaluate). See Uniswap-v3-Strategy-Backtest.
When getting a core pool instance by clientInstance.initCorePoolFromMainnet or clientInstance.recoverFromMainnetEventDBFile, Tuner has already automatically replayed events from the first record to the last one in endBlock. For replaying following events, you can load them by EventDBManager, and decide how to use them on your own.
An example of streaming process will be:
import {
ConfigurableCorePool,
EventDBManager,
EventType,
SimulationDataManager,
SimulatorClient,
SQLiteSimulationDataManager,
} from "@bella-defintech/uniswap-v3-simulator";
import { LiquidityEvent } from "@bella-defintech/uniswap-v3-simulator/dist/entity/LiquidityEvent";
import { SwapEvent } from "@bella-defintech/uniswap-v3-simulator/dist/entity/SwapEvent";
import JSBI from "jsbi";
​
// the database name containing downloaded-and-pre-processed mainnet events
let mainnetEventDBFilePath =
"events_0x8ad599c3A0ff1De082011EFDDc58f1908eb6e6D8.db";
​
// build a client instance
let simulationDataManager: SimulationDataManager =
await SQLiteSimulationDataManager.buildInstance(
"Your file path to save the internal data"
);
let clientInstance: SimulatorClient = new SimulatorClient(
simulationDataManager
);
​
// Specify an endBlock number
// the SimulatorClient will replay events up to that block
let endBlock0 = 12374077;
​
// get a pool instance
let configurableCorePool: ConfigurableCorePool =
await clientInstance.recoverFromMainnetEventDBFile(
mainnetEventDBFilePath,
endBlock0
);
​
// get an EventDBManager instance to load events
let eventDB = await EventDBManager.buildInstance(mainnetEventDBFilePath);
​
// get and sort event by block number
let events: (LiquidityEvent | SwapEvent)[] = [];
let startBlock = 1000,
endBlock = 2000;
let mintEvents: LiquidityEvent[] =
await eventDB.getLiquidityEventsByBlockNumber(
EventType.MINT,
startBlock,
endBlock
);
let burnEvents: LiquidityEvent[] =
await eventDB.getLiquidityEventsByBlockNumber(
EventType.BURN,
startBlock,
endBlock
);
let swapEvents: SwapEvent[] = await eventDB.getSwapEventsByBlockNumber(
startBlock,
endBlock
);
events.push(...mintEvents);
events.push(...burnEvents);
events.push(...swapEvents);
events.sort(function (a, b) {
return a.blockNumber == b.blockNumber
? a.logIndex - b.logIndex
: a.blockNumber - b.blockNumber;
});
​
// replay events
for (let index = 0; index < events.length; index++) {
// avoid stack overflow for the possible recovering operation
if (index % 1000 == 0) {
configurableCorePool.takeSnapshot("");
}
​
let event = events[index];
switch (event.type) {
case EventType.MINT:
await configurableCorePool.mint(
event.recipient,
event.tickLower,
event.tickUpper,
event.liquidity
);
break;
case EventType.BURN:
await configurableCorePool.burn(
event.msgSender,
event.tickLower,
event.tickUpper,
event.liquidity
);
break;
case EventType.SWAP:
let zeroForOne: boolean = JSBI.greaterThan(event.amount0, JSBI.BigInt(0))
? true
: false;
await configurableCorePool.swap(
zeroForOne,
event.amountSpecified
//,event.sqrt_price_x96
);
break;
default:
// @ts-ignore: ExhaustiveCheck
const exhaustiveCheck: never = event;
}
}
Note: Tuner doesn't export LiquidityEvent | SwapEvent directly but you can sitll use them. For external data(event logs), we recommend you to implement your EventDBManager to load and manage event logs, according to your context.