You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Looking for software to backtest/trade high-frequency algorithms
- Thread starter snackly
- Start date
Quote from snackly:
Well yes, you need to account for what affect you'll have on the market, however there are heuristics that one can use to approximate that. And of course that impact largely depends on the size of the order, the liquidity of the market, the depth of the market (presuming you're not in a multi-contributor market), and the exchange order priority algorithms, etc.
But thanks for pointing this out, it's very valid. That said, don't think that people don't backtest tick strategies. Clearly not everyone is crazy enough to just trade tick strategies without testing it against some set of assumptions.
I agree. I am also backtesting some high frequency strategy. I am not sure what set assumptions to use though.
For example, once the signal is generated, if I use the mean trade price in the next 1 second for buy/sell, I got fantastic backtest results.
But if I use the worst trade price in the next 1 second for buy/sell, I got extremely crappy results.
Any thoughts?
If you are using a market order, why not use other side of the bid/ask. Be sure to account for your order latency.
Limit orders are harder. You can get steped in front of and orders that should hit yours might trade on another exchange.
Your order can even be traded through due to top of book rules.
For limit orders, it is not a question of slippage, but if the trade happens at all.
Limit orders are harder. You can get steped in front of and orders that should hit yours might trade on another exchange.
Your order can even be traded through due to top of book rules.
For limit orders, it is not a question of slippage, but if the trade happens at all.
Quote from dloyer:
If you are using a market order, why not use other side of the bid/ask. Be sure to account for your order latency.
Limit orders are harder. You can get steped in front of and orders that should hit yours might trade on another exchange.
Your order can even be traded through due to top of book rules.
For limit orders, it is not a question of slippage, but if the trade happens at all.
Thanks. For the limit order, if the trade happens, does it guarantee a complete fill or just a partial fill?
And suppose by calculation we need to trade in a balanced manner x= 100 shares and y=200 shares, and the relation must hold, how do we do such kind of trades?