In the realm of algorithmic trading, there is an endless flood of statistics. Most of the software packages for backtesting provide huge amounts of data that can lead to endless (literally) analysis. Which ones are important depend very much on your trading objectives, style, preferences, and tolerance. This series has been dedicated to the data that I find important, or at least interesting, in the development of my trading systems. Today I wrap up the last of those data points I find useful. The system described in this series, the MultiStage v.6 Trading System, is the one I currently use to manage my Covestor.com model.
Win Ratio: This is the epitome of High Probability Trading. I hear that term thrown around a lot, yet I seldom hear a clear definition. In my view, I like to see something very near to 80% win ratio, and certainly greater than 75%, before I will call it high probability. The current parameters for the MultiStage v6 Trading System have produced about 79% over the past decade. Depending on the parameters I use when I backtest this system, the results are always between 77% and 81%.
Sharpe Ratio: If you don't know what a Sharpe Ratio is, this number is probably meaningless. If you do know what it is you should be impressed. The current parameters being traded using the MSv6 produce a Sharpe near 5. In simple terms, the Sharpe Ratio is a measure of excess risk-adjusted returns. In other words, it gets higher with better returns and lower with greater risk (i.e. volatility).
Sharpe likes nice smooth equity curves, low draw-downs, and excellent returns. According to Morningstar, " the average Sharpe ratio for all mid-cap growth funds is 0.29". If a fund manager can muster a Sharpe of 1.0 or 1.5, they are considered to be well above average. That should give you some idea of what I am striving for in system development.
Exposure: The next important data point I watch is exposure. This is an indication of how much time your money is in the market and how much time it is in cash. In the case of this decade long MSv6 backtest, exposure is about 24%. In other words, 76% of the time we are in cash. That doesn't mean we are never fully invested... many times we are. Conversely, much of the time we are 100% cash (like right now).
Maximum Positions Held: This is not really an output of system testing. Rather, it's an input. It's a statistic often asked about, however, so I thought I would include it. The number of positions held is always set to a maximum of twenty, but depending on position size, cash may be exhausted before 20 positions is hit. For example, if I put 10% into 20 positions I can buy 200% of my capital in positions, which works if I use all my margin. If I increase position size to 15% I will run out of cash before I hit 20 positions. That being said, position size is dynamic and determined by the system; so while 20 is the max it often isn't reachable even when fully invested.
I'll do one final post with a wrap-up of the results data, so stay tuned. Hopefully this series provides a good idea of what I watch and why it's important in the trading system development process, and while trading.
Good Trading...
No comments:
Post a Comment