Arbusers

This post is about discussing if it is true that odds are generally more overvalued for the home team than for the away team for the value bets we make. For the discussion of this subject, I will use my history of bets.

Period: dec 2017 up to now
Sport: soccer only
Types of bets: moneyline and handicap bets, full time and half time, pre-match only
Bookmakers: 2 offline books whose characteristics are similar to most soft bookmakers'
League sizes: approximately 20% top, 45% middle and 35% low

The yield and number of bets for the entire sample is as follows:
12,40 % (1248)

Now this sample is split into home team wins, the draw and away team wins:

Home team:
Full time and half-time: 19.01% (579)
Full time: 17.23% (531)
Half-time: 43.0% (48)

The draw:
Full time and half-time: 2.32% (113)
Full time: 1.85% (66)
Half-time: 2.94% (47)

Away team:
Full time and half-time: 6.50% (556)
Full time: 7.67% (534)
Half-time: -26.11% (22)

So the yield seems to be 3 times as high for bets on the home team than for the bets on the away team. I have checked the sample for the last 12 months only, and the tendency is the same, just on a smaller sample. Based on this I will start increasing the stakes on home teams and decreasing a bit on away teams.

So the question remains: If this is indeed a phenomenon for value bets being made, that home teams are more often being overvalued by the books than away teams, what could be the reasons?

You've probably considered this, but be careful changing anything right now.  Odds probably have not fully adjusted for the current no-spectator games we're seeing so the home field advantage could be lower than your older data suggests.

True Alfa1234. I won't be changing stakes drastically.

My sample since april 1st this year (full time and half-time):
All: 6,39% (289)

Home team: 15.55% (128)
The draw: -69,6% (18)
Away team: 4.26% (143)

I know the sample size is small, but from my sample I cannot see the effect of no-spectator games - the tendency is the same as before. Home team is better.

VidaBlue wrote: This post is about discussing if it is true that odds are generally more overvalued for the home team than for the away team for the value bets we make. For the discussion of this subject, I will use my history of bets.

Period: dec 2017 up to now
Sport: soccer only
Types of bets: moneyline and handicap bets, full time and half time, pre-match only
Bookmakers: 2 offline books whose characteristics are similar to most soft bookmakers'
League sizes: approximately 20% top, 45% middle and 35% low

The yield and number of bets for the entire sample is as follows:
12,40 % (1248)

Now this sample is split into home team wins, the draw and away team wins:

Home team:
Full time and half-time: 19.01% (579)
Full time: 17.23% (531)
Half-time: 43.0% (48)

The draw:
Full time and half-time: 2.32% (113)
Full time: 1.85% (66)
Half-time: 2.94% (47)

Away team:
Full time and half-time: 6.50% (556)
Full time: 7.67% (534)
Half-time: -26.11% (22)

So the yield seems to be 3 times as high for bets on the home team than for the bets on the away team. I have checked the sample for the last 12 months only, and the tendency is the same, just on a smaller sample. Based on this I will start increasing the stakes on home teams and decreasing a bit on away teams.

So the question remains: If this is indeed a phenomenon for value bets being made, that home teams are more often being overvalued by the books than away teams, what could be the reasons?

you need more matches. on soccer odd even market after 1000 matches.  I had 8% roi on odd and when reach 1680 matches both odd and even was negative

antemartic wrote: you need more matches. on soccer odd even market after 1000 matches.  I had 8% roi on odd and when reach 1680 matches both odd and even was negative

Earlier today I thought:

Are you sure? With such a high yield it should be near impossible to dip below zero again.

I almost answered something like this, but before I pushed the post button, I made a Monte Carlo simulation on my own very high yield bets just to further prove the point.

Confession: you're right, I need more matches. Variance had me fooled again, even when I thought it was no longer possible.

The result of the Monte Carlo simulation is attached.

So in the simulation I assumed EV 12.4% on the entire sample. The simulation did 10.000 runs. The probability distributions are well-known bell curves as expected. It was quite surprising that the high yield of the home matches is not even within the area of statistical significance in the traditional sense (beyond 2 sigma). I had me convinced that deviating more than 6 percentage points from the middle over more than 500 bets would be quite extraordinary. However, had I used the yield from the away sample, 6.50%, as the text book answer for EV, the home sample would have deviated very significantly.

Statistical significance, 2 sigma, is just a tradition, and we may not have to wait for this significance to appear or we may wait forever. If simulations show that some patterns are in the top 10 or top 20 of a probability distribution, maybe that is enough to start changing the stake size, simply based on the indication that it is not only variance?

So in your sample antemartic, it is likely that the sudden dip after 1000 matches is due to plain variance as you probably believe yourself. With a little more details of your sample, such as how many bets you had when yield was 8%, what the average odds were, etc., maybe we could simulate it to sort of quantify the possibility of other factors than variance being present.