I’ve been thinking about how we could dynamically capture the live spread size between the buy and sell price, and use this within trading strategies to avoid times of very high spreads. Either in real time, or to save the data into a database and use it to avoid times of high spreads.
When developing a system, the platform only gives us the mid price and we input our estimated spread. I was thinking if we record the executed buy price and then compare it to the next bar’s mid price, the difference could act as a proxy for the spread. By then offsetting this calculated spread against the rolling P&L, we might just be able to uncover the hidden cost of the spread in real time.
I’m aware this method isn’t without a few quirks, market noise and lag could muddy the numbers, and I’m thinking it’s highly probable that it will never work since we are only ever working one mid price in PRT and what we might end up recording the slippage cost rather than the spread cost. I’m just thinking of ways to get a reliable estimate from somewhere.
Before I go any deeper, has anyone else dabbled in similar ideas? Or are there alternative methods out there to capture live spread data? I’m thinking more for stocks, rather then indices, since indices tend to have fixed spreads for most of the day. Although it would be useful to record them into a database too.
.
