OHLCV Candle format

Trading Strategy has a specific OHLCV (open-high-low-close) candle format that is more tailored for decentralised markets. Trading Strategy uses (first trade, last trade) style candle pricing over traditional cryptocurrency open/close price where open is the last close.

• The opening price open is the price of the first trade during the candle period.

• high / low / close fields are the same as for normal candles.

• OHLV data sparse: if there are no trades, no zero volume candle is generated. If your strategy needs to have zero volume candles, you can forward fill the data when you are constructing the trading universe.

This is because we believe this formats fit better for DeFi market data feeds

• There is no information lost, it is only presented differently. More information is retained in this format.

• Candles are snapped to underlying blockchain block ranges: trades can only happen within a block and are physically represented by (block number, log index) tuples.

• On AMM markets, the price can only move if there is a trade event.

• There are no gaps between close and open because markets are 24h. Furthermore there are no opening and closing auctions as one might have at stock markets.

Forward filling data

Transforming Trading Strategy candle format to the traditional open/close definitions can be done with following:

• For strategies that read a candle open price, just choose to use previous candle close price instead.

• For strategies, be mindful that there might not be a candle, because there are no trades. Many of the DeFi markets are low volume, and might not see any trades for minutes and sometimes hours. Thus the timestamp of the last candle might not be necessarily the previous tick.

• Replace the open price with the close price of the previous candle.

• For any missing sparse timestamps, generate a candle where open/high/low close are set the the volume to zero using forward fill.

Some methods for forward fill your data:

• tradeexecutor.strategy.trading_strategy_universe.TradingStrategyUniverse.create_from_dataset() takes forward_fill argument (False by default). You can use this in your create_trading_universe function to forward fill loaded data.

• For low-level forward filling see tradingstrategy.utils.groupeduniverse.resample_candles() and tradingstrategy.utils.groupeduniverse.resample_series().

Distinction between buys and sells

We include logic to determine if a trade was “buy” or “sell”

• On AMM markets there are no true sides: everything is a swap. Sell is a swap with reverse token order compared to buy.

• We consider anything to be buy if the amount of quote token decreases after the trade (your USD goes down, ETH balance increases).

• See additional rules regarding the Determining quote token.

Large differences between close and open

Sometimes for exotic and illiquid pairs there might be a great different between the last close and open price. Some potential explanations incluse

• Mint / burn mechanisms of tokens

• Backdoored / scam tokens

Extra price feed columns

The following extra fields may be avaiable depending on the payload

• start_block

• end_block

• buys

• sells

• Volume might be split to separate buy_volume and sell_volume columns

For the details of the candle fields see tradingstrategy.candle.Candle.