Some critical reading on trading

The key question I have been working on for the past couple of months is this: how long does one need to monitor a strategy in order to gain confidence that the strategy produces the Sharpe ratio advertised? I had not yet been able to calculate the Sharpe ratio for the simulated data as I felt I was missing a key point about such calculations - something in the back of my mind told me that there was an answer to this overall question, but I just couldn't remember where I had seen it

Then I started reading Irene Aldridge's book High Frequency Trading , which I had bought a few week's back but delayed reading till now, and on p 59 is the precise question above and the method by which the relevant Sharpe Ratio is calculated. As I suspect, the Sharpe ratio from daily data has adjustments in it to annualise it, based on the same sort of change for re-basing volatility from annual, to monthly, to daily

Armed with this new knowledge I am able to calculate the annualised Sharpe ratio from the figures I have. Indeed, I had actually correctly calculated the daily Sharpe Ratio, my issue was solely how to convert this to an annualised number. And my answer - the calculated Sharpe ratio is 12.5, an extraordinary number at first sight, but not uncommon with high-frequency intra-day trading systems

Highly recommended

And the author herself

On the related question of how much testing is necessary, the high Sharpe ratio implies that relatively little data is needed to verify the figures - as little as 20 days if the claimed Sharpe ratio is high enough. So I perhaps have enough data to be confident that I am on the right tracks (at the very least)

This leads to another question and one that I have been thinking about since re-reading The Predictors book, which is the issue of speed of ramping up position. Clearly if you were sure your Sharpe Ratio was 12.5, you could ramp up very quickly. But I am being more cautious. I have a minimum opening position sizing and am hoping to ramp up four times before the end of the year, ending the year at perhaps 3x the position size I started at

I remember back in 2000 meeting a UBS macro hedge fund guy at their offices in Broadgate. He showed me the room where the Prediction Company's "black box" was actually sitting, chugging away on its own - fully automatic. He claimed this had a Sharpe ratio for over three years of trading of 16! Such a system is profitable virtually every hour - a claim mentioned at one point in Patterson's The Quants

.

What would a system look like that could be profitable every hour? Well the best analogy is perhaps with a casino running banks of fruit machines (hundreds). Even if each machine produces a big jackpot every hour, the entire operation would be profitable every hour - it is the making of thousands of small bets with the edge in your favour that produces this effect. So a trading system that is profitable every hour would have to trade thousands of times a hour with a tiny edge in its favour. Easy

More essential reading - read it a few week's back in New York, thinking of reading it again very soon.