Profitable, or fooled by Randomness?

Yes, I know I’ve nicked this title from Nassim Taleb’s eponymous book, which is worth reading, by the way. But especially in this case and in the text I lay out before you, I think it’s highly appropriate.

I’ve traded hundreds of thousands of markets on Betfair, most of which I have collected detailed data on, then archived and analysed. I use data for a lot of things now, but the main reason I first collected data was to get a base measure with which to trade against—in classical experimental terms, a control sample.

This blog post reveals why knowing that is the key to trading profitably.

Are you lucky or skilful?

It’s very easy to be fooled by randomness when you’re Betfair trading. The key concept here is to understand whether you have an edge or have just been lucky.

When trading, any skill that you thought you may have, may turn out just to be random fluctuations within whatever it was you were doing. If you don’t understand that concept have a watch of this video where I explain how I compare data to underlying metrics: –

Regression to the mean

It’s a bit like a striker in a football team that’s on form. Every shot they have seems to fly into the back of the net. The striker who is not in form has the opposite problem. Every ball they hit seems to be ricocheting off the defenders, off the post, over the bar. It does everything apart from hitting the back of the net.

Therefore, it’s easy to fall into the trap of assuming that somebody possesses or doesn’t possess skill when, in reality, it’s just part of the natural variation of whatever you’re doing. Eventually, both sides will regress to the mean.

Regression to the mean happens all over the place, and there is a great example in this article: –

https://www.spectator.co.uk/2011/12/he-knew-he-was-wrong/

Why should you collect data

I’ve gathered data since day one on Betfair. I use it to model markets, compare and contrast, and extract insight. But that’s only possible with vast amounts of data, and it takes time to achieve.

Back testing using commercially available data sets is available, but I’ve found them deeply lacking in insight—like looking at something in 2D rather than 3D. It helps to use real money and get real data on your participation in the market. So that’s what I’ve done for the near-on 20 years I’ve been in the market.

But another reason I collect data, and I suggest you should do the same, is so you can work out what would happen if you had no skill at all and just did things at random. That way, you can understand what influence you are having. If you can understand how your decisions influence outcomes, you will know if you profit in the long term, whatever temporary issues you have.

It’s easy to go on a good run and increase the stakes. Only to see your strategy collapse just after you did that. You need to find something to define what you are doing, which is not simply a positive moment within the market.

It’s a very easy trap to fall into. I’ve done it, you will have done it, and everybody does it at some point. But I’ve learned now, so I’ll always need a base measure within a market to understand if I am truly skilled.

Trading isn’t a 50/50 process.

If you are an active trader, you don’t need me to tell you that trading isn’t a 50/50 process. It’s not random because it is a coin flip as to whether you get it right or not. But if you are new to trading, you may not know what I’m talking about. Let’s have an example.

Let’s say you are Betfair trading and have chosen scalping within a horse racing market in the last five minutes before the off. If you choose just to do a single one tick scalp, this will exhibit a high win rate by default. This is because the market will quite likely fill an order offset by just one tick in this high liquidity period near the off.

Even if the market deviates from the current price, it’s quite likely to return to or past that price in the other direction before the start of the race.

Of course, trading is a balance of profits and losses, so if you have +1 strategy on upside and -20 on the downside then your win rate needs to be extraordinarily high to be profitable overall.

I did a nice video that explained strikes rates and probably which you can view below.

The three really important things you need to know

One of the very first things I did when I started trading way back in the year 2000, was to trade at random. I already knew that random wasn’t 50/50, the number changed depending on a number of variables. So I wanted to isolate that as quickly as possible.

This sounds counter-intuitive at first. If you trade at random surely it must be 50/50, but the answer is not in the input, but the output.

If you trade at random you will break even, but it’s a combination of strike rate, another name for how often you get a winning trade. This is then combined with your wins and losses combined. Put these together and you get your trading expectancy.

Random trading will produce varying strike rates depending on exactly what you are doing. You can pretty much make up your own strike rate. If you wanted me to show a string of 10 or 20 positive trades I could just pick a high strike rate strategy. It wouldn’t be profitable in the long term, but it would look impressive.

It’s trap!

This is a trap that most newbie Betfair traders fall into frequently. People, quite naturally, focus on a strategy that wins. Whoop, whoop, I’m winning!

But failure is ensured long term if all you are doing is trading at random, but with a high strike rate. If that rare loss is substantial, it will wipe out all your gains. This has nothing to do with being unlucky.

Ultimately though, that is not the key to successful trading, as the video pointed out. Being profitable in the long term is about getting those three key variables correct and in the right proportion.

Why in-play is different from pre-off

If you trade pre-off, you will find that, in the long term, winning and losing trades tend to be equally distributed. But in-play is a whole different ball game.

If we just lay a horse at a specific price, let’s say 5.00, then in an efficient market, we will win roughly 80% of the time. Boom! Well, not really, because that is just what you would expect given the odds. To profit, you need to exceed that 80% win rate, plus commission, to make any money longer term. But this isn’t a trade; it’s an outright bet. Let’s turn it into a trade.

Let do some scalping in-play instead. We are going to lay the favourite at odds of 5.00 and offset the trade by five ticks. Lay at 5.00, out at 5.50.

At odds of 5.00 there are plenty of ticks above the current price. By laying first we are effectively playing a game whereby our strike rate has to be at least 80%. That is because this horse is going to lose 80% of the time and our closing trade will be matched. Boom!

Therefore, we can say the win rate of this trade is at least 80%. But what about the 20% of the time that the horse goes on to win?

Because we are trading rather than just outright value betting, there will be times when the horse wins but doesn’t make all the running. This means that at some point during the race, its price will drift. We are not asking whether it will win 20% of the time. We are asking how far the price will move before it eventually wins.

As I’ve modelled the in-play market extensively, I can tell you this would happen 70% of the time at starting odds of 5.00. So, there is a 70% chance, 20% of the time, that you will still get your trade matched.

I won’t get into complicated maths at this point, but this trade would work 95% of the time. It’s almost certain that you would get this trade matched and be able to show an impressive string of results.

To put this in context, you can take that percentage, invert it to decimal odds, and effectively back at 1.05 shot. That’s the beauty of a betting exchange: you can flip and play with your options and represent them in a clearly defined manner.

At 1.05 shot is like backing a football team winning in the last few minutes of a football match. It will work most of the time, but that’s actually not what you are worried about.

Your focus shouldn’t really be on winning

But, of course, there is a catch in everything we have talked about so far. Ok, you win 94% of the time, but you should be more worried about that elusive 6% that won’t get matched, that comes in.

If you lay £100 to open your trade, your liability will be £400 or 4 times your stake. But of course, you have limited your upside to just £9 by trading in and out. Your liability, if it all goes wrong, is nearly 45 times your maximum win.

So, inverting that into a strike rate, means you need to win roughly 98% of the time, excluding commission, to break even with that mix of upside and downside. Basically, you need to win pretty much every time to have a chance of profiting in the long term. That’s unrealistic.

How you should approach any trading strategy

Knowing that there is a natural win rate in a market if you trade it at random, should tell you a lot about what your core objective is. You need to exceed that rate or else you are just being fooled by randomness.

Framing your trade around an entry and exit price and the profit that comes out of that, will tell you what strike rate you are looking for. Tweaking these numbers in effect creates your target win rate. All these things are interchangeable.

This makes high probability events much harder to identify whether you’re using actual skill. You will win for long periods and now and again be “unlucky”. Low probability events display the opposing characteristic and events near a central value sort the wheat from the chaff quite quickly.

The only way that you can confirm if you are profitable is to repeat your action over a very long period. That’s logical because over that period, if you exhibited no skill, you would eventually run out of money due to frictional costs.

As I have pointed out many times before, this is why a lot of trading advice is merchandised around high probability systems. It will frequently win; when you lose, it can be just discounted as bad luck. Something that maybe won’t happen next time around. But the reality is, the system’s or a process’s true strength is its longevity.

Therefore, in addition to strategies that run every day, I also have automation that sits alongside my standard trading activities. The objective is really simple: Let’s see what would happen if I used no skill.

The simple fact is that you need to know roughly what the underlying metric is within the market. Otherwise, you have no chance of exceeding it. Or if you’re going to get serious and quantify it, that will tell you exactly what your edge is within the market. 

Therefore, the first thing you should do if you are presented with any trading strategy is to understand what the underlying base rate is. Doing that will tell you if you are just lucky or whether they had some element of skill.