I first wrote about the underlying results this season in the
January monthly review. You can see the comments here.
As discussed in that monthly review, until this season, I had no
need to really track the underlying results for the leagues I was playing in.
Quite simply, when you are making money month after month, trying to understand
what drives your profitability isn’t high on the agenda. Of course, I was fully
aware that last season was a really good season for backing aways but
considering I’d made better profits the previous two seasons when the
underlying conditions weren’t quite so favourable for aways, I didn’t really
think too much about it.
Before we look at the results in depth, if we take a step back, my
thinking at this game has been pretty transparent from day one and anyone who
follows my bets will be aware of this. I am looking to go against the crowd and
my ratings tend to find bets that aren’t easy to find if you simply looked at
league tables, recent form and so on. All of my bets are mathematical based on
my ratings algorithms I’ve developed and therefore, if I was sitting down
without my ratings being updated each week, I would struggle to find all of the
bets I end up having. However, that’s not to say the bets are impossible to
find and even if I was guessing, I’d know the sort of teams I would be backing
each week. Anyone with the service could find many of my bets every week if
they understand the patterns involved.
I have always employed some parameters around my base rating
algorithms and over the first 3 seasons, these parameters have helped to boost
my returns. I’m not giving too much away here when I say my ratings struggle to
find value in short priced home bets and in short priced away bets. Therefore,
in games where there are short prices, I’ll either find no value in the
underdog or I’ll be backing the underdog. I won’t be backing the short-priced
favourite!
To be clear, that’s not to say that I don’t believe there is value
to be had in backing favourites. If I was given a choice of backing favourites
with 10% value or outsiders with 10% value, I’d choose favourites every time as
the strike rate would be much higher and the P&L journey would be much
smoother for me. However, my ratings are structured in such a way that finding
value in short prices is difficult and therefore, I’ve always been wary of
this.
My other issue which is intrinsically linked to the odds issue is
that my ratings tend to favour away bets. Again, it’s not something I set out
to achieve and if I was doing this again from scratch, I would probably try to
find a way to rebalance the rating algorithms but basically, 70%+ of my bets
are away bets. This season is the first season I’ve brought in Draw systems and
therefore, if anyone is following a wide range of systems, they can have Homes,
Aways and Draw bets in their portfolio but the bet mix will be weighted
according to the systems they follow.
So, putting this together then, my ratings are going to be finding
a lot of value in bigger priced away bets. By definition, my ratings must
therefore be opposing short-prices (both Home and Away) and therefore, this
gives an idea of the types of bets that anyone following the systems will be
having every season.
Where has it gone wrong this season for my ratings and the
systems?
Well, it’s actually quite easy to pinpoint the issue. Quite
simply, this season has been nothing like the last 3 seasons in terms of
underlying results and the places where my ratings tend to find bets have been
much more difficult than has been the case historically. That's not the only issue and I don't think my ratings have been as good this year as historically and ultimately, the ratings haven't had enough of the rub of the green this season but maybe they got this for the last 3 seasons!
Let’s drill down into the figures a little more for this season.
Here’s the results if blindly backing all outcomes this season so
far in the UK leagues I rate (top 5 English and SPL):
As you can see, at a high level, if backing all Aways this season,
you would have lost 7.6%. If backing all Homes, you would have been sitting
with a profit of 0.8%. Clearly, if you are picking twice as many bets from a
sample of bets that have lost 7.6% rather than a sample of bets that have made
a profit of 0.8%, then you are starting from a bad position. However, as
always, it’s never as easy as this!
As I said in the intro above, my ratings don’t find many short priced
home bets and they definitely don’t find any short priced away bets. Therefore,
to understand the impact the underlying results have had, we need to break down
the returns further.
Here’s the returns for Home bets split into 3 categories based on
odds:
As you can see, although blindly backing all home bets is slightly
better than break-even, if you narrow your bets to odds on teams, you can make
a return of 7.1% this season! An amazing statistic really and sort of explains
why I’ve read so much this season about bookmakers taking a beating when so
many odds on favourites have been winning. Unfortunately for me and my ratings,
I can count on one hand the number of odds on bets I have at home each month
and therefore, I’m likely to be backing the aways in these games!
When I do back home bets, I’m backing teams priced 2.00+ and more
often than not, 3.00+ and as you can see, the returns aren’t great but are
probably a little better than long-term expectations. My home bets are a little
bit worse than long-term expectations this season and therefore, I think it
highlights my ratings haven’t been as powerful as usual for these bets.
I think that’s an important point here. I’m not hiding from the
fact my ratings haven’t been as good as usual this season, I can see that from
the results. The point I’m making is that my ratings aren’t quite as bad as
they appear though as they are being badly impacted by trends in the underlying
results being different from usual.
If we flip this on its head then, what’s the returns of the away
bets split by 3 categories of odds:
Well, if Homes were an issue for my ratings, Aways are a bigger
issue. I can count on one hand the number of away bets I’ve had this season
priced 2.75 or lower and therefore, I haven’t been able to capitalise on the
fact that these bets have only lost 1.2% (which again, is lower than long-term).
All of my away bets fall into the range 2.75+ and unfortunately, this is where
the large losses have been this season. A loss of 7.5% for the bets priced up
to 6.00 and then a crazy loss of 24.1% for bets priced 6.00+.
Taking all away bets priced 2.75+, it’s a loss of 9.9%. Therefore,
I’m picking all my away bets from a sample of bets that have lost 10%. Even if
my away ratings have an edge, it’s going to be difficult to make a profit over
a very large number of bets.
It’s probably worth pointing out here that number of bets is
important here. If I am picking a small number of away bets from a sample of
1,465, with a little bit of skill and luck, I could hit a profitable set of
bets even if backing all bets loses 10%. My issue is that I’m backing a high %
of these bets given I have 4 algorithms on the go and multiple combined systems
which are picking the same bets over and over again and therefore, the more
bets you choose from the sample, the more difficult it is to make a profit.
Putting homes and aways together then and looking at bets priced
2.50 and below and over 2.50 (90%+ of my bets are over 2.50), we have the
following results:
I think this is the best indication of how difficult my ratings
have found this season. If backing teams greater than 2.50 this season, it has
been very, very difficult. A loss of 8.7%. Conversely, if backing favourites,
you would have found it difficult to lose this season although as my home bets
have shown, just because the conditions are easier, it doesn’t mean it’s easy
to make money!
I guess one thing anyone reading this might be thinking is, so
what? This is the name of the game and ultimately, these sort of things
level out in the long-run. Well, let’s take a look at the previous three
seasons for the same subsets of bets as well as this season:
As you can see, for the first two seasons, the losses for backing
all homes and aways were in line with the long-term expected losses of circa.
2%-3% for these odds ranges. My ratings and systems had two very good seasons
in both of these seasons and therefore, based on the long-term averages, I was
sure my systems and ratings would perform well long-term.
Last season, we had a bit of a freak season whereby backing bets
priced 2.51+ was very profitable and backing low priced teams was massively
unprofitable. Given my ratings play in the range that was very profitable, last
season was another very good season for my ratings and systems and as I
commented on lots of times last season, it was fairly easy to make a profit
last season using my ratings.
Looking at this now, this season is now nearly the complete
opposite of last season and ultimately, if it was easy to make a profit last
season using my ratings, it’s nearly impossible to do so this season.
Unfortunately, as I’ve come to realise this season, my ratings are
impacted by the underlying results and this season has shown me that even if I
have a profitable set of ratings long-term (or 4 sets of ratings in my case!),
I can still make a loss if the underlying results don’t fit in with the
long-term and if they are biased against my ratings. Last season didn’t fit in
with the long-term but fortunately, the were biased towards my ratings not
against my ratings and therefore, I wasn’t impacted.
Of course, one way to reduce this bias is to have a blended mix of
bets and to not only have larger priced bets or away bets and that’s definitely
something I need to think about for the future. Is it worth giving up a
little bit of ROI and ROC each season over the long-term in return for better
protection during the seasons that don’t perform like the long-term?
I think that’s the crux of this post. I have done lots of other
analysis to try to understand some of the trends were are seeing this season
and therefore, I’ll share the analysis below with some comments. Anyone
following the systems this season will probably be able to tell me all about
these trends without me having done the work as we experience them during the
season but it’s always interesting to see if your thinking fits in with the
data results.
Here are the underlying results by League:
Lots of observations I can make here but here are a couple that
make me chuckle:
- Bsq Prem Homes have been very difficult to make a profit from the underlying results historically and have never returned a profit in any season from 06/07 (as far back as my data goes) – showing a 12.7% profit this season. As you would expect, my ratings struggle to find too much value in these Home bets in this league!
- League Two Aways have been very difficult to makes profits from historically for my systems and have returned a profit once over a season in the last 6 seasons. A profit this season (although it's lower than what it was a few months back).
One other thing that has been apparent this season has been the
difficulties my ratings have faced with midweek games. Interestingly, I was a
little surprised at these results as I thought they would be much worse but I
think it points to my ratings underperforming for midweek games more than the
underlying results. No idea why and historically, I’ve had no issues with
midweek games on the systems but for whatever reason, this season has been
horrible for midweek games.
Anyway, here are the underlying results:
Clearly, I don’t have enough away bets on a Monday or Thursday
evening!
As you can see, Wednesdays have been horrible but there are so few
games on a Wednesday, it will just be variance. Most of my bets are on a
Tuesday!
I could break Tuesday’s down and show the fact that longer prices
in midweek matches this season have performed poorly but we already know this.
We’ve been betting on them!
The issue with day of the week is that the sample sizes are so
small apart from a Saturday, you can’t draw any real conclusions from the data.
However, you can see the short-term bias in the results and therefore, midweek
this season has been trickier than usual but I’m not likely to be dropping all
midweek bets going forward! My draw systems have done well on Tuesdays though. :)
In terms of performance by month this season, the figures look
like this:
I like this table as it paints a pretty good picture of my
struggles this season. September was a nightmare, January was really tough and
March was a nightmare. Looking at the Away ROI by month, you can sort of see my
struggles this season. Interestingly, December and February were good months for
my Away bets by the way. :)
In terms of the Home bets, September was a great month but
generally, my Home bets haven’t been as good as you would have hoped based on
the underlying results.
As some of you will have twigged by now, looking at home and away
results like this is difficult as when I’m saying that Aways underperformed as
more Homes were winning than usual, I can’t really expect my Home bets to do
well as I can’t be backing the Homes and Aways in the same game!
Clearly, although I’m saying it’s been easier this season to make
a profit backing Homes, this is the only the case if I’ve not been backing the
Aways. I’m not going to try to work this out as it’s too difficult to be honest
but you get the gist of the issue.
To finish this post off, here’s an interesting picture which shows
the results by month for my 4 rating algorithms since they went live and the
underlying results for each month also.
Again, you could spend all day looking at this but the point I’m
making is that there is a correlation to the returns from the algorithms and
the underlying results. Hopefully, a sign of a decent rating algorithm is that
it can outperform the underlying results long-term and therefore, you only have
to worry about short-term blips impacting the returns. IMO, a large part of
what we have seen this season is a short-term blip but I believe the ratings
need refreshed with more recent data and therefore, I’m holding my hands up for
the poor performance this season as well as saying that it has been more
difficult than normal.
To end this post, here’s a reminder of all live results for all
systems since inception. We can talk all day about underlying trends,
short-term blips, poor seasons, Home bias, Away bias and so on but the results
of the systems are the results of the systems. 52 systems, 45 in profit since
they went live, 6.5% profit for all UK bets. Not as good as it once was but
still not too bad in my opinion. Over 32,000 system bets over 4 seasons. I
haven’t fluked 45 profitable systems from 52 I hope!
No comments:
Post a Comment