I have an algorithm that makes one round trip per day that I would like to take live but I'm having trouble reconciling the performance over the last two months paper trading on my Interactive Brokers account vs the same time period in a backtest. Since the beginning of November the paper trading simulation has returned 13% and the backtest over the exact same time frame with the same code has returned 32%, quite a difference. The backtest assumes $.005/share $1 minimum commissions, leverage on both is under 1.0, long only. I've looked through the trades and they match between the two simulations within a minute or so and the number of shares always match, the only difference I really see is the fill price of the orders never matches. Obviously both are a simulation, I'm just hoping for feedback from people who have taken an algo live and what their experience was with respect to backtest vs paper trading vs live fill prices.
I implemented a vix daytrader algorithm which when deployed live also had results that varied from the backtester and even the contest results. I highly recommend calculating the realized P&L (accounting for commission and slippage) of every trade and then graphing the total P&L in your recorded variables rather than rely on the backtester return. Make sure to match the starting capital with your actual starting capital.
I wonder also if there's any comparison between backtesting with minute and tick data. Indeed fill prices and even getting in and out of positions can vary significantly at certain periods of time.
Have you seen this? How Accurate is Our Slippage Model: Comparing Real and Simulated Transaction Costs
Also, if you are trading low volume securities, have you seen this "bug"?
Thanks for the replies. This particular strategy trades fairly liquid securities, currently the 20 day sma of volume is over 30 million. I had not seen that article on slippage yet, it did provide some insight. I think the main source of the discrepancy is the fact that these securities move quickly so there could be a significant price difference in the minute between order and simulated fill. At this point I think I'll edit my algorithm to send limit orders rather than market orders as well as record/graph prices at the limit price thereby underestimating fill price and ultimately returns.
The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian.
In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.