Rowleyfile Investigates: The Not-So-Dreaded Timeform Squiggle

There was plenty of constructive feedback to the initial Rowleyfile Investigates about Timeform squiggles [available here] – I may be contacted on Twitter at @RowleyfileTF or through @Timeform1948 – to the degree that a follow-up seems to be required.

One correspondent quite correctly asked whether Timeform squiggles did what they were supposed to do in predicting underperformance and/or a shortfall in wins. Without proof of that, other assumptions are of questionable validity.

With that in mind, I decided to try to answer a couple of basic questions about Timeform squiggles, using the same dataset of Ireland/UK Flat racing 2011 to 2013 inclusive.

Do Timeform Squiggles Predict Underperformance?

Yes, they do.

Horses with pre-race squiggles gained 435 wins where 502 could be expected at random, for an Impact Value of 0.87; they beat 47.8% of their rivals, where 50% is par; and they performed 2.6 lb worse on average compared to their pre-race Timeform Master Rating than horses without squiggles.

Interestingly, their performance was particularly poor in Ireland (IV of 0.68, 44.3% of rivals beaten) and was worse on turf (0.80 IV, 47.0% of rivals beaten) than on all-weather. Squiggled females (0.72 IV 45.8% of rivals beaten) fared worse than their male counterparts.

Does The Market Adjust For The Underperformance of Squiggle Horses?

Yes, it does: in fact, it overadjusts for it.

Betfair SP predicted 400 wins (the sum of 1/BSP for all cases) for squiggle horses. As we have seen, 435 occurred, so that is an actual over expected (A/E) of 1.09. This A/E rose to 1.19 for horses running at mid-range distances (8f to 11.9f) and to 1.17 for all-weather compared to just 1.02 for turf.

In simplified terms, in a 10-runner race where horses might be otherwise expected to have an equal chance and all be priced up at 10.0, horses with squiggles should be priced up at 11.5 in recognition of their likely underperformance, but they are priced up at 12.5. The degree of this mispricing varies according to circumstance but exists for all categories with a reasonable sample size.

Do not expect squiggle horses to win more often than chance – the opposite happens, which is a relief for the reputation of squiggles – but do expect the market to pay such horses too little respect.

It is fair to assume that this was not the fate expected of squiggles when they were devised, but a lot has changed since that time, including, presumably, the public perception of squiggles themselves.

If you are prepared to accept a lower strike-rate but a bigger potential reward, then squiggles do identify underperformance but are not-so-dreaded as the betting market seems to think they are.