Tuesday, 2 June 2015

2014/15 Season Underlying Results

I’m now in review mode for the 2014/15 season. I’m actually not looking forward to it really and after the season I’ve had, reviewing what has happened isn’t going to be much fun but it can’t be good all the time. I was probably due a poor season after 4 really good seasons although the season before last didn’t actually feel like that great a season either. 

The last 2 seasons now have been below the profits experienced in the first 3 seasons and therefore, I suspect the systems and ratings have a lot of questions to answer going forward. When I refer to this, I’m really referring to the UK systems as we know them. The Euro ratings & systems have had a second successive profitable year and the UK draws did OK this season, so it’s not all doom and gloom. I won't mention the Euro Draws!

Anyway, the various systems and ratings will be reviewed over the next few weeks.

I thought I should start this summer with a review of the underlying results. I’ll deal with the UK results first and then cover the Euros later on in the post.

If we can recap for a moment, it’s worth reiterating why it’s worth looking at the underlying results. Historically, there is a clear correlation between the underlying results and the results of my ratings. If we see fewer Home wins in the season and more Aways for example, I would expect my ratings to possible see fewer Home wins and more Aways. If the strike rate for Aways is very high, I would hope my ratings would catch more Away wins and therefore, we’d potentially make more profits on Aways. Of course, the opposite is true for Homes in this circumstance and my ratings would probably struggle with Home bets if we see an Away bias.

Of course, nothing is ever perfectly correlated and we have seen spells before where there was a spike in Homes or Aways and my ratings haven’t caught any of the bias at all. If we’re backing lots of Homes bets and there is an Away bias, it’s not much good to us if we’re backing the wrong teams. Likewise, sometimes the bias could be from short-priced Aways and because my ratings and systems filter out all low priced away bets, we don’t catch any of the bias and we’ll usually be on the Home bets in these games, thus creating large losses.

Generally, I have used the underlying results to try to understand the fluctuations I am seeing in the results in the short-term. Over the course of the season, things tend to revert close to the mean we would expect although every season, there will always be a slight bias towards one or two of the outcomes. Given I have systems which look at Homes, Aways and Draws, my aim was always that by taking a diversified approach over a season, we could somehow always ensure that we didn’t end up with too many eggs in one basket.

Saying all of the above, this season, I have been aware of the fact that for my UK ratings, the P&L hasn’t been in line with the underlying results. Even from my investigations, I’m struggling to see why my ratings and systems haven’t had a better season in the UK leagues and therefore, it is a bit of a head scratcher. Usually, with these underlying results, I would have expected a decent season and therefore, it again points to something being a little bit different this season than the previous 4 seasons.

Here’s the underlying results this season alongside the last 4 seasons:

Overall, the Home ROI is the best it has been since I started out 5 seasons ago with a loss of only 0.6%. The Aways have had their second best season since I started and it’s amazing to think that following all Aways created a profit of 2.9%. The downside from these results is that Draws had their toughest season and a loss of 6.8% if backing all draws. The strike rate for draws was only 25.3% over the season which is the lowest strike rate since 2006/07 season.

So, putting these results together, I would have bet pretty big bucks that my UK rating algorithms (systems 6,21,31 & 41) would have a profitable season (maybe as good as any of the previous 4 seasons) and my Draw rating algorithms would have a nightmare season and a big loss. I was very impressed that my draw algorithms made a profit the previous season when the draw strike rate was lower than historically, so I would expect the draw algorithms to really struggle this season with an even lower strike rate.

As you can see from the picture above at the right hand side, the results for my ratings were nothing like what I expected to happen.

In arguably the second easiest season (in terms of underlying results for Homes/Aways) since I started, 3 of my 4 UK algorithms have made a loss over the season. In easily the hardest season for UK draws since 2006/07, both my UK draw algorithms have made a profit. My second draw algorithm somehow made a profit of 5.8% if backing all games on this algorithm. Considering the underlying results were a loss of 6.8% (this is to best prices remember whereas the P&L I’m quoting is to Pinnacle prices only on a Thursday evening), I would have to say that the performance by system D6 this season is maybe the most impressive performance from any algorithm since I started.

On the other hand, I would have to say that my 4 UK algorithms have under-performed massively looking at these results this season.

As I said above, it is a bit of a head scratcher and it shows us that simply looking at the high level underlying results and expecting our P&L to follow suit every season is never going to happen.

Here’s the full breakdown by month since I started:

If we look at the months for the past season, we can see that for the first 2 months of the season, unless you were backing Homes, you had no chance of making a profit. As we know, Homes are what my ratings have the least of and therefore, it’s no surprise that my algorithms struggled to make much headway these months. The 4 UK algorithms were pretty much around the break-even mark and the UK Draw algorithms had dug themselves into a substantial hole looking at the ROIs here.

November was a slightly easier month and amazingly, the loss on Homes, Aways and Draws was very similar across each outcome. Surprisingly, system D1 had a great month and system D6 had a nightmare month. Funny how a month which had a loss of 1.8% for backing all Draws can see a profit of 25% on algorithm 1 and a loss of 17% on algorithm 2!

I think December is another tricky month to fathom. An amazing month for Aways, a nightmare month for Homes & Draws. I would have expected my UK algorithms to make a profit and the Draws to have a nightmare. 3 of the 4 algorithms had a losing month and amazingly, the draw algorithm which had a nightmare in November had an amazing December.  A profit of 18.4% when the underlying results lost 12.2% is an amazing result.

January underlying results were pretty good and my 4 UK algorithms made massive profits. The best month of the season by far for the ratings. Another poor month for the UK Draw algorithms and it’s been a tough season so far for Draws!

February should have been a dream month with an 8% and 9% return for Homes & Aways but it just didn’t happen for my ratings at all. Algorithm 1 made a 19.4% return but algorithm 3 made a loss in one of the easiest months the ratings have had in terms of underlying results since I started. Another loss for Draw ratings this month which was to be expected.

March is easily the month where I get very, very confused. A return of 24.8% if backing all Aways. Same story as December, 3 of the 4 algorithms have made a loss this month. Before this season, it is hard to find a losing algorithm in a month where the Away bias is so favourable and yet, it has happened twice in one season! Draws were a substantial loss again given the low strike rate.

April then came along, was very favourable for Draws, a nightmare for Homes & Aways and boom, the 4 UK rating algorithms run up a massive loss in the month. The worst month the ratings have ever had together in total and considering they hadn’t made good profits during the favourable months previously, no surprise to see the way the season turned out for these ratings. Although the month was favourable for Draw bettors, nothing like what the Draw ratings managed to do and the algorithms achieved a 20% and 45% profit to recoup most of the seasons losses in one fell swoop!

May was again favourable for Homes/Aways and although 3 of the 4 algorithms capitalised, the bet numbers were small and it didn’t make too much of a dent in the overall results for the season. Algorithm 2 somehow made a loss too this month which means for 3 very favourable months for Aways, the ratings haven’t capitalised at all. The Draw ratings ended the season well although the bet number was tiny.

So, in a season that was very favourable for Homes/Aways, the ratings have had a nightmare and whilst I would have expected losses for my Draw ratings, they did very well!  All in all, a very strange season I think.

Here’s the breakdown by league for this season:

I always think a season isn’t a long time for looking at this sort of breakdown and if you look at the historical results, the breakdown by league does bob around every season.

The Championship was a league I struggled with most of the season to get my head around and looking at these results, I sort of see why. I never seemed to hit the right Home/Away winners in this league all season and the draw was a pest.

The SPL is another league I struggle with and looking at the underlying results, it’s disappointing that this is the case. Again, I can’t seem to hit the correct home and away winners in this league.

Here’s the breakdown by month again which is maybe easier to see than the picture above:

Here’s the breakdown by day of the week:

Although all of the above is interesting, I’m not sure any of it explains why we’ve had a tough season this season. If anything, it definitely points to this season not being any tougher than the last 4 seasons on average and therefore, it’s disappointing my ratings have struggled so badly.

The last place to look is the odds. We know that from the above, backing all Homes/Aways was profitable this season. We know that my ratings play in the higher odds ranges and therefore, the last question we need to answer is related to the odds. Has there been a bias towards longer prices this season?

Here’s a pic simply showing the P&L if backing teams above/below 2.75:

Given about 90% of my bets come from the group 2.76+, it is so, so frustrating to see that it has actually been a profitable set of bets.

To recap, the same results last season is shown below:

As you can see, the previous season was a much less favourable year and my ratings did pretty well. This season is probably the easiest year I’ve had since inception and my ratings have had a nightmare.

Here’s a breakdown of the odds ranges for the 2014/15 season split by Home and Aways:

It starts to look even worse!  We know that most of my Aways are in the 2.75-6.00 odds range and backing all these bets returned a profit of 6.5%. Therefore, for me to not have a good season after seeing these underlying results is mindboggling to me!

Right, how do I summarise this?  Well, quite simply, all 4 of my UK rating algorithms have had a very, very poor season. The worst season they’ve had since inception and given how favourable the underlying results appear to be, you have to really question my ratings after last season.

On the other hand, my UK draw ratings appear to have had quite an exceptional year. Yes, the profits aren’t great but given how tough a season it was for backing UK draws, you have to think that when the strike rate increases a little for draws, these draw ratings can only get better.

I’m so used to using the underlying results to show that my ratings are outperforming expectations but for whatever reason, this happened for the 2014/15 season. The UK Draw ratings have done very well but all 4 UK algorithms underperformed badly.

I’ll quickly look at the Euro underlying results now. 

Here’s the results by season updated to include last season:

At a high level, these numbers look favourable for all 3 outcomes last season. Homes are probably a little down on historical experience (remember there has only been 2 losing seasons for backing all Home bets since 2007/08) but even so, there was still a marginal profit if backing all Home bets in these leagues.

Aways were a marginal loss and again, looking at the historical results, slightly worse than the last two seasons but much better than previous seasons.

Draws had their 2nd best season out of the last 5 seasons, both in terms of strike rate and ROI although it was still a loss making season. As I picked up in the pre-season post last Summer about the Draws in these leagues, every season for the last 9 seasons have now been loss making for the underlying draws and with an average loss of 7%, backing draws in these leagues seems a difficult place to be. Of course, the upside of this is that backing Homes (and possibly Aways) is the place to be and I think the results I’ve seen this season for the Euro systems back this up.

I think when you see the fact that the Underlying draws only lost 2.7% and yet, my Euro Draw ratings lost 9.8% and 13.6% respectively, it backs up what we picked up very early on in the season and that these ratings were rubbish. I guess I wasn’t totally surprised although as always, it’s disappointing to spend time building something to basically watch it never get off the ground or a better metaphor is probably getting off the ground and then crashing and burning very quickly.

On the other hand, the Euro ratings have followed up their first good season with another great season and I would have to say that based on the last two seasons, these ratings would probably be my best ratings. Considering the first attempt I had at building Euro ratings failed miserably, in a weird way, it sort of gives me hope for the Euro draw systems if I want to have another go at producing a new set of ratings.

I guess the biggest difference between having a second attempt at Euro Draw ratings and my second attempt at the Euro systems was that I was always confident I could find an edge with a set of Euro ratings. The underlying results look OK and I’m sure I just needed a second attempt at building the ratings again and I would have a good shot at getting it right.

After seeing the strike rate for Euro Draws again this season and seeing the fact that all of the underlying profits this season have come from one league (Italy), it doesn’t exactly fill me with confidence that trying to build another set of Euro Draw ratings is the way to go. The other 3 leagues made massive losses for Draws again and therefore, I have to think that if the bookmakers are going to continue to under-price draws, backing Homes/Aways is the place to go.

In terms of the Euro ratings, I have to say I’ve been massively impressed with these ratings over the last two seasons. It’s a shame I didn’t nail these ratings first time round as their proofed results are ruined a little by the losses in the first season but even so, the last two seasons have been impressive.

Algorithm 1 (System E1) has had 2 losing months in the last 14 months and for any rating algorithm, this is impressive stuff. The second algorithm has had a few small losses which has meant a little bit of red has crept onto the page but even so, the results are decent.

The actual season results show that algorithm 2 is slightly more profitable than algorithm 1 over the last 2 seasons but there isn’t much in it.

Overall, I’m looking forward to seeing if these Euro algorithms can build on the last two seasons and if they can, I suspect the focus will switch away from UK bets and more towards Euro bets for most of us following the systems.

Here’s the breakdown by league:

This is what makes this game so interesting at times for me. A few people have suggested I have different ratings for different leagues based on historical trends but the truth is, the results over a season are purely random. No one knew that Italy was going to be a goldmine for Draws any more than Germany would be a goldmine for Homes. Based on the previous 4 seasons, Italy was the worst place to be for Draws and my ratings had picked this up clearly. I had fewer Draws in Italy and boom, an amazing season comes along for Draws in Italy.

Historically, France was the best league for Draws and although the average odds were lower in this league, there profits in 2 of the last 4 seasons for underlying draws in France. Boom, an 18.4% loss for underlying draws in France and of course, my ratings took a bath here.

Backing all Homes in Spain had been profitable for 7 seasons in a row before last season. Boom, a 6.7% loss this season. Italian Homes has been the other place to be historically, profits in most seasons. Boom, an 8% loss last season.

As I said, this is what makes this game interesting and if it was as easy as simply following historical trends, everyone would be making money at football betting.

My Draw ratings were clearly not smart enough to get around these changes in trends this season and unfortunately, they suffered the consequences as a result. On the other hand, my Euro ratings have navigated through the season well and didn’t get too caught up in these short-term blips.

Here’s the P&L by month again in a table that’s easier to read for this season:

Here’s the P&L by day of the week:

Interestingly, one thing I noticed the season before last was that there was a slight Home bias on a Saturday and a Sunday and a slight Away bias in midweek. I have noticed the exact same trend last season. Not sure I’d start acting on this but it will be interesting to see if it continues next season too.

For Euro bets, most of my bets fall into the 2.51+ price range rather than below this. The results have been slightly favourable in this group last season.
Here’s the breakdown between Homes & Aways for a few odds ranges.

Big priced Homes and big priced Aways was the place to be last season generally and thankfully, this is where a lot of my bets occur in the Euro leagues.

To summarise the Euro results then, a very good performance from the Euro rating algorithms against favourable conditions although to be fair, the conditions appear favourable every year for Homes/Aways in these leagues. Euro Draws underperformed massively and the results were a little more favourable than normal. However, finding any sort of edge in these leagues with Draws looks to be difficult to me and I guess I need to seriously consider whether or not it is worth the effort of trying to find an edge. 

No comments:

Post a Comment