Football Betting - Football Results - Free Bets
Updated: 17/05/24 legal disclaimer
Acca Boost


Football-Data's revenues come from losses accrued by customers of advertised bookmakers. Nearly all bettors lose in the long run. Users taking advantage of any advertised welcome offer should familiarise themselves with the bookmaker's T&Cs. Any information appearing on this website, including references to winning, profit, beating odds/bookmakers etc. should not be considered professional advice or a recommendation to bet or gamble. Furthermore, the small proportion of customers considered by bookmakers to be sufficiently skilled at beating their prices may find their stakes restricted or have their accounts closed. Both the UK Government and bookmakers alike maintain that betting and gambling in the UK should be considered a form of entertainment only and not a trade to make money. Please gamble responsibly. NEVER risk what you cannot affort to lose. GambleAware.

Steamers and Drifters in a Football Match Betting Market

Posted 11th February 2016

In my second book How to Find a Black Cat in a Coal Cellar: The Truth about Sports Tipsters, I provided some analysis looking at whether studying the movement of betting odds can in anyway offer predictive clues as to the outcome of a football match. According to the efficient market hypothesis, betting prices that hold positive value expectation will be exploited until that value no longer exists. In layman's terms, this means prices shorten when people bet them, and lengthen when they don't. It is the simple process of supply and demand. Consequently, one can reason that the direction in which a betting price is moving can provide clues as to whether the original betting price held value or not.

Strict disciples of the efficient market hypothesis would argue that any price inefficiency will disappear the moment knowledge about it hits the market. In practical terms this would mean almost nobody will be able to exploit it. However, the analysis published in my book suggested something else. Discounting the effects of the bookmaker's profit margin, English football match betting odds that had been shortening (or steaming) prior to kick-off had in fact not shortened (or steamed) enough. Conversely, odds that had been lengthening (or drifting) had not sufficiently drifted based on an a posteriori analysis of results. That is to say, discounting for the bookmaker's profit margin, steamers and drifters still held positive expected value at the time of market closure.

This analysis was based on 2 seasons worth of data, specifically the 2010/11 and 2011/12 seasons for the 4 professional English football divisions. With a further full 3 season's of data now available, I decided it was a good time to revisit this work to see if it had all just been a lucky occurrence. As for the original analysis, again I used the average prices for the home-draw-away market as reported by Betbrain (for pre-closing odds) and Oddsportal (for closing odds). Given the difference in data sources, I must again recognise this as a shortcoming. Nevertheless, as originally reported, the bookmakers used to compute the market averages are largely the same across both odds comparison websites, and as for the original analysis, the average price and overround for each source was almost identical (3.178 and 107.1% for Betbrain versus 3.189 and 107.2% for Oddsportal).

The degree to which betting odds steamed was calculated simply by computing the ratio of pre-closing price (Betbrain) to closing price (Oddsportal). Hence prices which steamed would always have a ratio above 1.00 (the highest being 1.358). For drifters, the reverse ratio was calculated, i.e. closing price/pre-closing price, so as to again ensure that all ratios were above 1.00 (the highest being 1.365) and symmetrical with the steamers. Standard deviation was 0.0342 (or 3.42%). The total dataset contained 10,180 matches and therefore 30,540 individual home, draw and away betting prices. Of these, 14,107 prices drifted, 14,387 prices steamed and 2,046 stayed the same. The tables below categorises the steamers and drifters by the strength of the price movements, more than 5% change, 2.5 to 5% change and 0 to 2.5% change, and shows their corresponding returns to level stakes.


Percentage price change (PPC)CountAverage pre-closing oddsAverage closing oddsPre-Closing price yieldClosing price yield
PPC > 5%3,0533.653.98-18.49%-11.20%
2.5% < PPC ≤ +5%2,9883.393.51-18.50%-15.63%
0% < PPC ≤ +2.5%8,0663.153.18-9.41%-8.38%


Percentage price change (PPC)CountAverage pre-closing oddsAverage closing oddsPre-Closing price yieldClosing price yield
PPC > 5%2,6593.433.15+7.09%-1.41%
2.5% < PPC ≤ +5%3,1473.032.93+1.78%-1.77%
0% < PPC ≤ +2.5%8,5813.032.99-5.40%-6.47%

Unexpectedly, the pre-closing prices of the steamers contained far more value than those for drifters. That, of course, is how it should be, since returns to shorter/longer prices will be smaller/larger. Indeed, the biggest steams were actually offering positive value expectation, since their betting returns were profitable. The more interesting observation comes from the returns of the closing prices. If this market was operating according to strict principles of market efficiency, those percentage returns should all be equal and roughly equivalent to the average market overround. But they are not. Returns from closing prices for the biggest drifters are still double digits negatives, whilst those for the biggest steamers are just barely below zero. In other words, the drifters didn't do enough drifting whilst the steamers didn't do enough steaming. This difference is highly statistically significant (p-value = 0.0002).

The next table summarises all steamers and drifters, and compares them to the full sample (including those prices which stayed the same).

CategoryCountAverage pre-closing oddsAverage closing oddsPre-Closing price yieldClosing price yield

Of course, none of this actually helps us tell which the biggest steamers are going to be before they start steaming. Nonetheless, the analysis does make it clear that prices appear not to achieve full efficiency by kick-off.

Just how consistent has this inefficiency been over the 5 seasons? The next tables make it clear.


SeasonCountPre-Closing price yieldClosing price yield


SeasonCountPre-Closing price yieldClosing price yield

We can really only make the case that in one season - 2013/14 steamers - did prices move by market closure to values that would imply something close to efficiency. Indeed, on this occasion they may even have overshot the mark. In general, however, a very consistent pattern of insufficient steaming and drifting persists across the seasons.

Given the slightly higher prices for drifters versus steamers, there remains the possibility, as I reported in my book, that some of the underperformance of drifters relative to steamers might arise from the favourite-longshot bias. The simplest way to test for this is to remove the bookmaker's margin according to the odds model I have described in previous articles. The next table reports the returns to fair pre-closing versus closing prices for drifters and steamers alike.

CategoryCountAverage pre-closing oddsAverage closing oddsPre-Closing price yieldClosing price yield

Clearly, the closing market inefficiency is still present. The difference in returns from closing prices between steamers and drifters is still statistically significant (p-value = 0.002), and the conclusions must remain the same: prices which move significantly appear not to move sufficiently far enough to truly reflect the "true" probabilities of outcomes.

Whether this inefficiency is sufficient to overcome the bookmaker's margin is another matter. Had one chosen to bet the closing market average prices for all 597 selections over this 5-season period that witnessed an odds shortening of more than 10%, one would have made a 7.5% profit over turnover. I've estimated that the probability such a performance might happen by chance, assuming expectation to be a loss of -7% on turnover, to be about 2%. That's statistically significant at the 95% level, but not at the 99% level. My personal view would be that this sample size is too small to be sure such profitability would be consistent. However, we are not restricted to betting market averages, and why would we choose to? Unfortunately, one must appreciate the likelihood of having betting accounts restricted if we target those bookmakers who offer prices way in excess of the market average, particularly if other brands have already shortened their odds. Outlier prices are outliers for a reasons - to attract new customers. But if you abuse such generosity on a regular basis, don't expect to be active at such bookmakers for very long.