Lately I've been looking to do more analysis of my results. I think I've admitted in the past, that this hasn't been a strong point of mine. In the middle of April I decided to ask Betfair for a spreadsheet of all the tennis matches I had traded since the start of 2007, so that I could catch up. Since then, I have kept my own sheets with every match in it, the volumes I traded, average odds and match stats too. In that time, I've traded 200 matches, and the analysis of this is becoming very worthwhile, with some pleasing and not so pleasing results.
So what are the numbers that matter? And why?
You want to work out the average of your wins and losses. The mean of all your results. Clearly this is showing you what a match is worth on average, the sum of all your work is in this number, the bigger the better.
Next is your average win, and your average loss. When you win, how much on average is that? And the same for when you lose. Now compare these two numbers, are your wins on average bigger than your losses? If you lose on average more than you win then it's clearly going to be quite a battle, I know this from experience!
Now work out the percentage of markets you win on. The size doesn't matter, just look at where you won and lost and work out a %. This number, combined with the ave. win and ave. loss determines how well you are doing, and will describe to some degree how well your strategy is working, and maybe what type of strategy you are using.
For example, you might be someone who scalps a lot of trades or backs short prices where your win % is very high, but your average loss is much bigger than your average win. Or perhaps you don't win often, but when you do, the payoff is much bigger than the loss.
You have to juggle these numbers and get them in the right ratio in order to win with your strategy. If you are a trader, it describes something of the balancing act you are trying to perform. (Ideally, this analysis would be made on individual trades and not overall market results... if you keep detailed enough records to do such analysis, then that is excellent, I don't, yet...it gets complicated with uneven matched amounts etc) Though it's not an exact representation, your win % effectively shows how often you are right, but you will need to run your winners longer and cut your losses quicker than than this %.
For example, if you make 5% on an ave. win, or lose 5% on an ave. loss, then the % of times you are 'right' needs to be over 50% to make a profit.
My own figures show I lose slightly more on average than I win, but crucially, the % of markets I win on is 60%, so I'm ahead. This is not the way I would like it however, emotionally, it is much easier to average more on a win than on a loss, it can often feel like 2 steps forward, 3 back, but a 60% win ratio keeps it going in the right direction.
Emotional stability is important for maintaining discipline. Keeping records like this allows you to actually see a bigger picture of the war you are fighting rather than the losses you might take in individual battles - which inevitably stick in your mind longer than the wins.
It's this emotional side of things that can wear a trader down over time, we're all different in how we handle losses, some push on and are able to accept things, others suffer negative pangs on the back of each loss, causing further damage down the line. This is a good point to introduce another statistical measure - variance. More accurately, standard deviation!
Many people will remember this from school with some dread, it's worthwhile doing. It gives you a measure of a set of results around the mean.
Typically around two thirds of your results will fall within one standard deviation. In simple terms it's a display of steadiness or consistency of results. If you are one of those people uncomfortable with large swings in fortune, then you need to work on keeping this number lower. Regardless of how low you keep this, there will always be some randomness to the path your profits and losses take.
Here is a superb resource from vertical solutions called the P&L Forecaster. You enter your average trade, result, week or month etc and the standard deviation of those results and it produces a 'forecast' for the next 100 measures of your unit. Eg, you work out your average week, and the standard deviation of all your weekly results and enter these numbers, it forecasts for the next 100 weeks.
It's important to note that this is simply a measure of how varied your next 100 results could be. Try the same figures again and you will get a different looking chart each time, because of the variance in your P/L. Play around with it a bit, enter in some high standard deviations, and some low, notice how up and down a high variance set of results is and how smooth a low standard deviation chart looks.
This should be telling you how important the bigger picture is and how random you can expect your results to be. It helps to understand that losses (and wins) are part of the picture, you can't get down too much when you lose and high when you win. As Brett Steenbarger says, "chance alone will affect the paths of returns. A trader who understands that it's not just about returns, but risk-adjusted returns, can best adapt to these trading realities."
And so we return to the smaller picture. The interesting thing I think to be learnt from this is that the macro image is determined by the micro details and actions we take when trading. A series of results where the win is bigger than the average loss, is down to individual trades with good upsides and small downsides (which happened to be more likely to occur than the market thought).
Getting an extra 0.5% on an opening bet or closing bet can change the variance in your P/L, managing positions better in terms of risk and reward has the same effect. As does selectivity and patience. Cutting losses ultimately reduces the size of your ave. loss, running winners the opposite effect to your ave. win, both reduce the need to be 'right' so often.
So, everything is interlinked down to the fine detail in trading terms, but our mindset and emotions need to be centered on the macro scale of things, understanding how each individual trade, market, week and month, affects the bigger statistical calculations. Some things I have found most rewarding from this analysis:
- Cutting losses quickly reduces my average loss, and in turn makes improving the bigger picture a lot easier to achieve.
- Taking profits in order to 'get back on' later, (I call this my churn rate) has reduced my own variance slightly, by not looking for such big payoffs, which in turn require more time and potential for losses. Result: ave. wins and ave. losses are both slightly smaller, so standard deviation has dropped... My P/L Forecaster chart has been smoothed out slightly and perhaps I'm less on edge that I might be.
- Waiting for more premium risk/reward opportunities has helped get the averages closer to one another or even favouring the win side since I did my analysis.
Most people still don't keep records of their activity, without records you can't really see where you are going wrong. You might not even know you are going wrong, which is the major hazard with 'mental accounting'. Perhaps this is a topic for another post.. for now I'll leave you with a couple of my figures, my win % is 60%, and the standard deviation of my wins and losses 3 days ago was 5018, now down to 4701.