I am sure some of you remember that a long time ago I promised you a review of a betting product known as Trademate. The Trademate team has given me access to the product and I have been tirelessly gathering tones of betting data ever since. Having collected a record of more than 11,500 resolved bets I decided the time to write the review of Trademate has come.
Luckily, as I was preparing to write this article I got a message from the Trademate team inviting everyone on the platform to a stream where they discussed the aggregate results of the tips the platform was producing. This was a great opportunity to compare the data I have gathered with the overall performance of all the platform’s clients. It has helped me confirm some of my observations, but also get a new perspective on my results.
An important point I would like to make is that after writing the first version of the article a major issue with the settlement of tennis bets was discovered, which severely affected my results. After settling the problematic bets manually I have updated the results and my comments on them.
Before I dive into the analysis, let me give you a short overview on what the review will be about. I will of course share my experience with using the product, the functions that the software is offering and the customer support. And while all that is certainly important, the million dollar question is whether the product delivers on its promise to find an edge on the betting markets and improve your betting returns. Therefore I will focus most of the article on answering this question. There were some things I liked and some things I believe could be worked on. I will make sure to go through all of these. Here we go.
Trademate is a web-based betting app that scrapes data from the betting markets in order to identify value bets and deliver them to its customers. How does it do that? It will take a base bookmaker offering the highest liquidity for a given betting market and compare all the others to it. Taking that difference and adding up the margin of the bookmaker, it will calculate the supposed edge for that selection and deliver it to you.
You can watch some videos from the Trademate Youtube channel to get a feeling of the platform. I have found it very intuitive and easy to use. You can build all kinds of presets to test different betting strategies. You can choose the bookmakers you want, the target edge, a desired range of time left till the start of the event and a few other things. The Core version gives you a bunch of soft bookmakers to choose from and with the Pro version you get three sharp books added to the list – SBO, IBC (Maxbet) and ISN. You get a section with your open and settled trades which you can use for bookkeeping purposes.
I have used the preset building option to build two main presets. The first one was the widest possible preset including all the books, odds, edges, etc. The goal there was to collect the biggest pool of data possible and cut it into smaller pieces later for a more detailed analysis. Then my second preset was one with only the sharp books, whereby I have added a sound alert to it so that I can catch as many of those as I could. I was particularly interested in the performance of this subset since this was basically what the Pro version offers on top of the Core one.
I have gathered a total of 11554 bets of which 3263 were with sharp books. Below are the overall results for the two subsets of bookmakers:
You will find a lot of data here so I will give you a quick rundown on what is what. Above you see two numbers for Total EV (Expected Value) and Closing EV. These are generated based on the assumed edge of your bet at the time of placing it and at the start of the event respectively. The closing edge is calculated based on the closing price at the base book. The colors seem to be a bit messed up so look at the labels.
You will see that the average closing EV always stands below the total EV. This means that the price at the base book on average moves closer towards the price we have taken as the event approaches. That is not all that surprising given that different books giving different odds represents market inefficiency and those tend to get corrected by the market. However it is worth noting that the gap closes from both sides – albeit surely not at the same rate – meaning that odds at the base book are not necessarily fully efficient. I have seen that in practice too in the form of many drifts at sharp books I have been victim of during arbitrage betting.
This brings me back to an exchange we’ve had with a reader in the comment section of my last article regarding what edge you have when betting the soft side of an arbitrage pair. As I said back then I think the base book is not fully efficient at the time the arbitrage is placed and I believe the numbers above confirm this.
Yet it is commonly stated that the closing price at the base book is the most efficient price you can get for an event. Therefore Trademate calculates a closing edge based on that. The closing edge is supposed to be the actual edge you have for this bet, not to be confused with the edge you see when placing the bet, as the closing price is supposed to be the most efficient one. Again, the margin of the base book is added in the calculation.
Therefore, if all is well, in the long run we would expect our return on capital to approximate our closing edge and our profits to stay around the Closing EV line.
Analysis of the overall results
In an earlier version of this article the profit line was running way below the closing EV line. The main reason behind this was that the data provider of Trademate seems to settle Over/Under Tennis bets wrongly. All of those bets were settled as losses whereas of course some of them were winners.
Overall the profit line runs closely to the closing EV line for a large number of bets. From this it can be concluded that the software does its job.
In my trading I have not aimed at maximizing my profit but at expanding the data set I have available for analysis. Using Trademate’s advice and my own intuition I have tested some more refined subsets in order to try and improve my results. In doing so I have noticed some interesting trends and also some issues with edge calculation within specific subsets. I will come to that again later.
What to bet on?
Before going to the subsets where it doesn’t go that well, let’s first see where we can find the most profit potential. In the help section of the platform the Trademate team recommends to look for higher edges the further away the start of the event lays. The logic behind this is that the variance of your edge is proportional to the time left before the start of the event, therefore from risk/reward perspective you would want a higher return to take on a higher variance. That makes sense.
Furthermore, in general the team recommends focusing on lower odds and betting closer to the start of the event. I believe that suggestion comes from their analysis of the community results where they have noticed that lower odds & betting closer to kick-off deliver better results.
I have to admit I was a bit baffled by that suggestion. While I understand the risk/reward point, I don’t get why would anyone drop high odds and early bets out completely. Yes, the risk for both is higher but there you can surely find value bets too. In fact, as it comes to time before kickoff I would expect to find more value bets earlier while the market hasn’t matured yet. Furthermore, concerning odds, the higher risk of high odds is considered by the Kelly stake formula, which is used by Trademate, so this shouldn’t be an issue at all.
I decided to test those suggestions using the data I have gathered.
Let’s start with the most obvious one. Higher edge must bring higher returns – I guess that doesn’t need further explaining. Let’s see how the picture looks for different edge ranges. Below I have put the results for 5 different minimum edges in the range of 0-8%.
The profitability in terms of average ROI does seem to improve with increasing edge. Somewhat similar effect can be observed for flat ROI, although the effect is not that strong there, which is a bit concerning. In general it might be concluded that higher edge will give you better profitability. Still, you get a negative flat ROI even for some high edge ranges, which can raise some concerns. I believe there is a clear indicator of value in those edge ranges as for higher edges I beat the closing odds with a higher margin, so the increasing profits are not the result of chance. Then what’s the matter with the flat ROI?
Have a closer look at the average odds for the different edge ranges. As we increase the edge, the average odds of the tips the platform returns increase too. Does that mean that the longer odds contain more value? I don’t think so. I believe that the edge might be overestimated for the long odds. This leads to the fact that for the flat strategy you get long odds selections, the value of which is in fact not that high, if there is a value at all. The Kelly stakes mitigate that negative effect as they adjust the stake to the odds but one can still see the issue in the flat ROI.
That aside, betting with a high edge will give you better profitability, but of course also a lower total number of bets. The advantage you have against the closing odds at the base book is a clear evidence for that. However, I think some “false positives” slip into the data set. I will look further into these with an analysis of different odds ranges.
I believe the task for everyone using Trademate would be to find a balance between a high enough number of bets in order to turn over the whole bank, together with a satisfying expected return per bet. I will give my suggestion on that later.
As I said, the Trademate team recommends avoiding long odds bets. Yet, according to the analysis above, long odds bets on average give you a higher edge, which to me is strange. I decided to look into how long and short odds are performing within the data set. I have looked into the results with minimum odds of 1-6.
The results confirm the suspicion that something is wrong with the long odds bets. Once your odds go above 4 most profitability metrics start deteriorating. I have repeated the test for all possible edge ranges and with very few exceptions the trend holds. It is telling that the flat ROI goes down as well for the high odds ranges. For me this is another clue suggesting that the edge for high odds might be estimated incorrectly.
So far, so good. We have established that higher edges give us higher returns, showing that the software does identify value. We have also noticed, strangely enough, that we should rather drop the high odds as something’s wrong there.
The Trademate team suggested focusing on bets short before kickoff to minimize your variance. Let us see how the performance differs across the Time-Before-Kickoff ranges. I have looked into bets placed 0-6 hours, 6-12 hours and 12-48 hours before kickoff.
The suggestion does seem to have some merit. We get slightly better returns and lower variance for the range of up to 6 hours before kickoff. However, the difference is not that big and I believe excluding the early markets altogether would rob you of too many good bets. If you are aiming at a lower total number of bets it might be a good solution for you, otherwise I would not recommend it.
I still don’t believe that there is more value to be found closer to kickoff. Intuitively it just does not make sense – this is when the markets are supposed to be most efficient. It is indeed the case that betting further away from kickoff increases variance in the edge and the results prove this. I have talked with the Trademate team about that and at this point I am sure there is no error in the margin calculation for earlier bets. My belief is that even though value bets do exist in the early markets, it is harder for Trademate to identify them as Pinnacle is not that efficient before closing. I guess using fundamental knowledge instead of technical methods like the one applied by Trademate, the sharp bettor might find just as many if not more value bets in the early markets as opposed to the late ones.
A Good Strategy
I was thinking about how an optimal strategy with Trademate would look like. We have seen that the profitability improves for higher edge ranges, that long odds bets suggested by Trademate are generally less profitable and that lower time-before-kickoff does seem to improve results somewhat but not by much. As I would like to aim at the highest total number of bets at a decent profitability I would go with an edge of at least 4% and will avoid odds higher than 4. Here are the results of such strategy.
The profit line moves next to the Closing EV as it should be. You get a solid profitability and a total of 1790 bets for around a month and a half of betting. Very nice!
You have perhaps noticed I have also restricted the maximum edge at 10%. The reason I have done that is that apparently the majority of bets with edge higher than 10% were palpable errors. As you probably know these will not only fail to deliver any profit (as they will most probably be declared void by the bookie) but will also limit your account in no time, so it is best to avoid them. Of course, you can include the highest edges too and decide on the spot whether it is a palpable error or not. Odds comparison website like oddsportal will help you for that. In my experience the majority of them are just that so be careful.
As I mentioned earlier, the Pro version adds three additional sharp bookmakers to the list – SBObet, IBC (currently Maxbet) and ISNbet. You also get Dafabet but I don’t count it separately as it shares odds with Maxbet. I am running at a small profit so far after a total of 3263 bets. However, that does not tell much by itself as in the overall subset I don’t manage to beat the closing line on average so I am still a bit skeptical. Therefore I have had a look at what happens when I increase the edge. You can see the results for the overall sharp data set as well as a minimum edge set at 2, 4 and 6%.
The profitability does look better by some measures as the edge is being increased. The edge against the closing odds increases. However, one can also see that the closing EV line is further away from the EV line at placement. It seems beating the Asians is much harder task as the Pinny odds drift much more in this case. The edges are slimmer and the variance higher. My sample is not large enough to determine what is the actual edge, but the community seems to be running at a 0.6% long-term ROI in the sharp books, which could be taken as benchmark.
Of course, given that we are talking about sharp books here the amount of bets you get is certainly smaller than with the softs, however I don’t think this would be an issue since you do not get limited, so you don’t have to focus on a league or sport and could just bet on everything. The average edge is much smaller in comparison to the whole data set, so you should not expect profitability as high as with the softs, but that’s not unusual either.
All in all so far I think the Pro package would be a valuable tool for the high-roller. I would strongly advise you, given the reduced profitability compared to the softs and the higher price of the product, to consider carefully whether you have the needed capital to make it work. Furthermore, considering the tiny edge you will have pursuing such a strategy, be ready for some serious variance down the road. You can refer to my article on betting variance to get an idea.
The Edge Calculation Issue
Alright, we have established that the software works and you can make a solid profit from it. But I am also suspecting that there might be a problem with the edge calculation for higher odds. What could the problem be?
A common trait for long odds bets is that the bookmakers apply to them higher margins than usual. I have been assured that Trademate does apply the margin when calculating the odds. However how exactly it is applied is unclear to me.
What is the case in practice? Intuitively, one might expect the margin for an event to be applied to all betting outcomes proportionally to their implied probabilities. However, this is not the case. In fact, longer odds receive an over-proportional share of the margin. It could be speculated what is the reason for that, but perhaps bookmakers want to increase their expected profit per long odds bet due to the higher riskiness such a bet brings them compared to a short odds one. You can read more on that in Joseph Buchdahl’s Fixed Odds Sports Betting: Statistical Forecasting and Risk Management.
Furthermore, margins vary per sport, league and country. In general, the margin can be easily calculated for any given event but it must be split between the different outcomes carefully. I might be wrong, but I think a further look into this can significantly improve the performance of high-odds bets in the Trademate platform. This would bring value to the customers as it would expand the total range of viable betting opportunities.
I haven’t said much about the support from the team yet. I would like to thank Marius and the rest of the team as they have been very responsive with every question I had. They also organize periodic meetings to discuss the results of the whole community. I really like that as it shows great transparency, which is extremely important in the betting industry. At those events the customers get the opportunity to see how everyone else is doing and to share their thoughts and feelings about the product with the whole community. All in all I have found the customer support to be at a good level.
Trademate does deliver on its promise to identify profitable opportunities on the sports betting markets. The team has developed an innovative software solution which would be an asset for any serious bettor.
I believe there are some flaws with the edge calculation of certain subsets that I have outlined above, which I am sure will be addressed by the team soon enough. Keeping those in mind you should have no problem turning a profit with Trademate relatively quickly.
Another point of improvement is the collection of data – there is currently an issue with the settlement of tennis trades in the platform, which has seriously impacted my records. I needed to settle those manually, which for data set this big was quite a lot of work. I hope the team will take care of that in the near future.
That aside, the software is easy to use and offers a lot of nice features. The customer support is on a very good level. Overall I would recommend the software to anyone who follows value betting strategies and has the necessary budget.
Some final words
This is certainly my longest article so far, so thanks for reading to everyone who has made it to the end. Thanks to the Trademate team for giving me access to their product, which made this review possible. I would also like to use the opportunity to thank Steve from Daily25 for giving a great feedback to my previous article and helping me improve the count of my Twitter followers substantially. Steve is offering a product based on Trademate for the Australian market, so if you are an Australian bettor make sure to check it out.
Also, if you would like to see any further stats from my data set in Trademate or want to ask anything just drop a comment or get in touch. I am sure the Trademate team will also be around to answer your questions.
On an unrelated note, I have prepared a short e-book on Sports Arbitrage, which you can download from the ‘Sports Arbitrage Guide’ tab on the top of this page. In that way you will also be added to an email list where I will update you on any new content in my blog. Alternatively, you can also subscribe to my RSS feed below this article or in the upper right corner. Thanks for following!
EDIT (12.03.2017, 12:39 CET)
Just wanted to let you know that I have presented the review to Marius from Trademate and he had a few comments on it. I am posting them with some delay here.
Concerning the edge calculation, Marius shared my opinion that there is an inconsistency in the margin attribution for long odds bets. He has let me know that the team is working on developing a solution. I am glad that the issue is being addressed by Trademate and I am sure that solving this problem will further improve the quality of the product. Marius has also stressed that it is advised to focus on lower odds in order to reduce overall variance, since many customers find out they are uncomfortable with the risk that higher odds bets carry.
Regarding the time before kickoff, Marius raised the valid point that the margin could be (and is) easily calculated using the odds at time of placing the bet. Therefore it is safe to assume that the margin is calculated correctly there. That fact only makes the underpeformance of early bets more interesting – since the margin is not the reason, there must be some other factor depressing the returns. At this point I am struggling to see what that might be, so if you have any ideas I would be happy to hear them. Regardless, this might turn out to be quite a useful finding by itself. It is good that platforms like Trademate collect large amounts of betting data so we can notice such particularities and improve our betting knowledge.
It was confirmed that all the closing edges are calculated against the Pinnacle closing line.
I have also added the missing link to the picture in the Time-Before-Kickoff Analysis part.
EDIT2 (26.03.2017, 18:56 CET)
It was discovered an issue with wrongly resolved tennis results has badly affected my results. It turns out all Total Game Tennis bets were settled as losses, which depressed the returns of my data set and most subsets significantly. Since the effect of this is substantial I decided to resolve these bets manually and update the article accordingly. I apologize to all readers for this confusion.