Then I deeply apologize. It is not apparent from your resume/public profile. Which software platform did you use to auto quote if I may ask? I am still highly confused by the statements you made re algorithmic optionstrading, in particular you said : "IMO, I would be careful automating option trading. WIth 15 option markets and unpredictable option spreads and liquidity at each price level, I find for a non-MM that manual executions provide a better results." Why is the number options markets relevant as long as one trades through a broker api and adheres to market specific parameters? Unpredictable spreads? All market variables are completely dynamic and unpredictable (though probabilistically estimable) in all asset classes. Liquidity can and does also vanish in other asset classes. I auto quote options and my stance is that as long as a solid risk model is in place and as long as one does not exceed the capacity of the strategy in any given market then an algorithmic approach to executing options orders is not any different from other asset classes. Of course are there asset class specific metrics that differ but those can all be accounted for in an algorithmic trading model.
If anyone is interested to run this system against their own IB paper trading account, I can share the compiled trading engine JAR (restricted to IB simulated account) and provide a VPN connection to my analytics server. Feel free to PM me...
Read the thread, still not clear on your fill assumptions. Is your backtest profitable even if you cross the spread on every trade (buy at ask, sell at bid)? You said you have data going back to July plus 1 year. That is not a very long backtest in this business. Sharpe of 7 is high, but you have to be careful of sharpe ratios in options trading, sharpe ratio can be borderline meaningless for non-normal return distributions. (I am not sure what the distribution should be like for your particular strategy, so ignore this if it's normal) But even if it's a legit 7 sharpe it could just mean you are well tuned to market conditions on such a short backtest. Additionally, sharpe can't actually be calculated from the data you posted, so we don't know if you compute it correctly. You mentioned panic selling in your real account, to me this is a sign that you are trading way too much size compared to your confidence in the system. IMO, automated systems need to be left to run their course unless you are intricately familiar with how your edge actually operates and you have a legitimately rational belief that your manual action has an edge over the system. So my advice is scale it wayyy down and let it run for a while, win or lose.