Females v Males. Is There A Bias Against Either Sex On The Racetrack?

Are female racehorses at a disadvantage when racing against males?

I did some research on this subject following a discussion on the horse racing forum and I was surprised to find that although the gender bias is widely acknowledged by the racing fraternity, it doesn't appear to have as much of an impact on the way punters bet.

Or at least, a check of past results showed that the betting markets might not be giving this bias enough weighting. 

Perhaps headline mares such as Zenyatta, Black Caviar, Moonlight Cloud, Treve and NH stars like Annie Power and Quevega have the effect of diminishing the view amongst punters that female racehorses are disadvantaged against males.

I'm sure there has been plenty of research done by systems bettors and others looking for a winner finding edge though. 

There was an article published on Flatstats.com in 2002 that showed a definite bias in favour of males and other Internet betting forums have touched on the subject.

The perception is that female racehorse are particularly disadvantaged on the All-Weather surfaces, and this is where my research took me after seeing the breakdown of the figures in Table1.

I've presented the results of the research in tables. 

The first part is a general look at all mixed-sex races.

Table 1 firstly has all races combined (NH Rules, Turf, and All-Weather), and then these figures are broken down into the 3 race types. 

It was at this point that I decided to limit the research to All-Weather racing for this blog.

Part 1.

The starting point was to look at all races which had the exact same number (50%) of Female and Male runners. 

The first figures in Table 1 are for every race combined, with no filters for Race Type, Age Groups, Distances, etc. 

Therefore the first set of figures will include all Mixed Sex 2yo Nurseries with 6 runners for example, alongside all 24 runner Handicap Chases, and every race in between. 

The only criteria for this first set of figures was that every race consisted of both Male and Female runners.

The Table then goes on to break down the 50:50 races into each Type of race (National Hunt, Turf, and All Weather).

To support the figures from the 50:50 races, the Table then continues with races that have a different ratio of Males to Females. I have looked at:

• All races that have from 1-49% males.
• All races that have from 51-99% males.
• All races that have from 33-67% males.

Again, each set of figures is then broken down into the 3 Race Types.

Table 1 probably covers more than it needs to. Each set of figures from the different ratios of Males:Females supports the figures from the first part (the 50:50 races) of the Table. 

The sets of figures for even more exact percentages of Males (exactly 33%, exactly 40% etc) would have smaller than ideal sample sizes, and showing them would be an overload of data, if it isn't already.

Table # 1 is only a very general overview. 

The figures suggest a marked difference in the performance of Female horses on the All Weather surfaces (and also in the National Hunt sphere), compared to the Turf, and I've mainly shown these figures to justify the focussing of my research on the All-Weather surfaces. 

There could well be certain categories of Turf racing that favours one sex over the other, but as with NH racing any research on the Turf will be done at another time.

The second part of Table 1 looks at the performance of the (outright) SP favourite according to the sex of the horse, from all mixed sex races that consisted of 25-75% Males (so in other words, no race had less than 25% representation of either sex),

The assumption was that a 2.00 favourite had a 50% chance of winning the race (disregarding the market overround) regardless of the sex of the horse. 

Plenty of research has been done to show that the finishing position of a horse correlates with its position in the betting market, and any bias should show up when looking at how the favourite performed, and whether or not the market has made adjustments to the price of any runner based on any perceived advantage to either sex. 

For this part of the table I've included Backing and Laying calculations, with Profit/Loss figures and ROI percentages at Industry SP, and Laying calculations at Betfair SP. 

These calculations didn't seem of any real benefit for the first section of the table, and weren't included there.

Race Type	Sex	Runners	% of Runners	Races	Wins	% Of total wins	A:E figure
Race Type	Sex	Runners	% of Runners	Races	Wins	% Of total wins	A:E figure
Race Type	Sex	Runners	% of Runners	Races	Wins	% Of total wins	A:E figure
Race Type	Sex	Runners	% of Runners	Races	Wins	% Of total wins	A:E figure
Table 1.
Races With Exactly 50% Males/Females
All Race Types.	Males	21108	50%	4374	2561	58.55%	0.88
	Females	21108	50%	4374	1821	41.63%	0.78
NH Races Only	Males	3532	50%	647	418	64.61%	0.88
	Females	3532	50%	647	230	35.55%	0.66
Turf Races Only	Males	11139	50%	2341	1271	54.29%	0.85
	Females	11139	50%	2341	1075	45.92%	0.84
AW Races Only	Males	6437	50%	1386	872	62.69%	0.93
	Females	6437	50%	1386	516	37.23%	0.74
The three parts of this table that follow probably don't offer anything of much value. They are included to give support to the first part of the table, and to give a general look at female performances from larger sample sizes. More useful figures are in the later tables that cover Age Groups, Distances, Tracks and Surfaces etc.
Races With 1-49% Males.
All Race Types	Males	36716	37.32%	9176	4050	44.03%	0.86
	Females	61665	62.68%	9176	5151	56.14%	0.78
NH Races Only	Males	3886	40.59%	786	452	57.50%	0.88
	Females	5688	59.41%	786	334	42.50%	0.63
Turf Races Only	Males	22805	36.95%	5615	2294	40.85%	0.83
	Females	38920	63.05%	5615	3331	59.32%	0.80
AW Races Only	Males	10025	37.02%	2775	1294	46.63%	0.93
	Females	17057	62.98%	2775	1486	53.55%	0.79
Races With 51-99% Males.
All Race Types	Males	690674	77.00%	78607	64655	82.25%	0.83
	Females	206784	23.00%	78607	14105	17.94%	0.76
NH Races Only	Males	312852	79.80%	32492	27963	86.06%	0.82
	Females	79674	20.30%	32492	4568	14.06%	0.71
Turf Races Only	Males	254353	75.20%	30077	23651	78.63%	0.84
	Females	84043	24.80%	30077	6496	21.60%	0.81
AW Races Only	Males	123479	74.10%	16038	13041	81.31%	0.87
	Females	43067	25.90%	16038	3041	18.96%	0.74
Races With 33-67% Males.
All Race Types	Males	190820	55.30%	31461	19918	63.31%	0.85
	Females	154356	44.70%	31461	11604	36.88%	0.76
NH Races Only	Males	47538	58.58%	6624	4749	71.69%	0.83
	Females	33652	41.44%	6624	1884	28.44%	0.65
Turf Races Only	Males	94335	54.03%	16005	9406	58.77%	0.83
	Females	80253	45.97%	16005	6632	41.44%	0.81
AW Races Only	Males	48947	54.75%	8832	5763	65.25%	0.90
	Females	40451	45.25%	8832	3088	34.96%	0.75

Performance of SP Favourite as Male/Female In Mixed Sex Races With No Less Than 25% Representation Of Either Sex.
Race Type	Sex	Runners	Wins	Win %	PL @ SP	ROI	P/L Lay	ROI Lay	A:E Figure
NH, Turf, AW combined.	Males	31,257	10,912	34.91%	-1,687	-5.04%	-558.32	-1.35%	0.95
	Females	12,543	3,835	30.57%	-1,433	-11.43%	-32.45	-0.17	0.89
NH Rules	Males	9,071	3,363	37.07%	-493.88	-5.44%	-169.28	-1.67%	0.95
	Females	2,262	687	30.37%	-364.63	-16.12%	+10.19	+0.36%	0.86
Turf.	Males	13,915	4,669	33.55%	-788.15	-5.66%	-60.48	-0.32%	0.94
	Females	6,972	2,158	30.95%	-603.22	-8.65%	-87.89	-0.83%	0.92
All Weather.	Males	8,271	2,880	34.82%	-405.05	-4.90%	-328.56	-2.62%	0.95
	Females	3,309	990	29.92%	-465.80	-14.07%	+45.01	+0.82%	0.86

Part 2. All Weather Racing.

The tables in Part 2 cover All Weather racing only. I've tried to find areas where Fillies and Mares may be at a bigger disadvantage than what the figures in Table 1 would indicate.

When breaking down larger sets of figures the problem of unacceptably small sample sizes comes up. I've kept each category fairly general in order to keep the sample sizes at a reasonable level, however it was unavoidable to include sample sizes of less than 1,000 for some categories.

There are two sets of figures for each category. The first set of figures covers every runner, and the second set looks at the performance of the outright SP favourite when either Male or Female.

I've only looked at races that had at least a 25% representation of each sex. This was mainly to avoid the situation of sole females competing against a field full of males, and to have the playing field as level as possible. The trade off for this was having smaller sample sizes.

The Laying calculations used for the SP favourite sections of each table include Betfair commission of 5%.

All Weather Racing. Females v Males.
Tracks and Surfaces.
Surfaces.All Runners.
Surface.	Sex	Runners	% Of Runners	Races	Wins.	% Of Total Wins.	A:E Figure.
Polytrack/Fibresand. Combined.	Males	17542	58.80%	12651	8659	68.45%	0.89
	Females	52897	41.20%	12651	4014	31.73%	0.75
Polytrack.	Males	62644	58.73%	10440	7143	68.42%	0.89
	Females	44102	41.27%	10440	3312	31.72%	0.74
Fibresand (Southwell).	Males	12777	59.21%	2211	1516	68.57%	0.90
	Females	8795	40.79%	2211	702	31.75%	0.79
SurfacesPerformance Of SP Favourite.
Surface.	Sex	Runners	Wins	Win %	PL @ SP	ROI	P/L Lay	ROI Lay	A:E Figure
Polytrack/Fibresand Combined.	Males	8229	2863	34.79%	-416.05	-5.06%	-312.47	-2.41%	0.95
	Females	3290	985	29.94%	-461.21	-14.02%	+45.62	+0.84%	0.86
Polytrack	Males	6749	2317	34.33%	-393.63	-5.83%	-254.73	-2.37%	0.95
	Females	2745	799	29.11%	-432.39	-15.75%	+71.18	+1.49%	0.84
Fibresand.	Males	1481	546	36.87%	-23.42	-1.59%	-56.79	-3.34%	0.98
	Females	546	186	34.07%	-29.82	-5.46%	-24.61	-3.63%	0.94
Tracks.All Runners.
Track.	Sex	Runners	% Of Runners	Races	Wins.	% Of Total Wins.	A:E Figure.
Kempton	Males	13332	59.00%	2206	1494	67.72%	0.88
	Females	9217	41.00%	2206	716	32.28%	0.77
Lingfield.	Males	18311	58.97%	3101	2200	70.94%	0.91
	Females	12606	41.03%	3101	906	29.22%	0.72
Wolverhampton.	Males	24407	59.00%	4161	2765	66.45%	0.88
	Females	17201	41.00%	4161	1401	33.67%	0.80
Dundalk.	Males	6334	56.00%	920	647	70.33%	0.88
	Females	4893	43.00%	920	274	29.78%	0.56
Tracks.Performance Of SP Favourite.
Track	Sex	Runners	Wins	Win %	PL @ SP	ROI	P/L Lay	ROI Lay	A:E Figure
Kempton	Males	1428	482	33.75%	-34.72	-2.43%	-119.32	-3.94%	0.97
	Females	564	142	25.18%	-138.80	-24.61%	+75.01	+5.87%	0.75
Lingfield	Males	2031	727	35.80%	-60.78	-2.99%	-77.55	-3.2%	0.96
	Females	753	242	32.14%	-58.97	-7.83%	-19.85	-2.03%	0.92
Wolverhampton	Males	2659	916	34.45%	-213.25	-8.02%	-58.88	-1.62%	0.93
	Females	1105	337	30.50%	-163.43	-14.79%	-4.68	-0.29%	0.85
Dundalk	Males	546	161	29.49%	-79.37	-14.54%	+2.50	+0.17%	0.88
	Females	282	62	21.99%	-77.64	-27.53%	+34.18	+4.16%	0.70
Southwell.	See Fibresand
Summary of Tracks & Surface figures.. I was expecting to find a bias towards either the Polytrack or the Fibresand, but the figures for both surfaces are almost identical. The breakdown of the tracks that use the Polytrack surface don't seem to reveal any bias against a particular track either, with figures for all tracks very similar. Females seem to have a poor record at Dundalk, but that could possibly be attributed to different race conditions for Fillies and Mares in Ireland, rather than the track surface, which is considered one of the better All-Weather surfaces. Favourites, in general, had a lower than average record at Dundalk. The green figures for laying female favourites at Kempton correlate with the below-average record of favourites in general at that track (the figures for males favourites at Kempton were also below Lingfield and Wolves). With Kempton and Dundalk accounting for the green figures for laying female favourites on the Polytrack, my conclusion is that nothing much of value can be taken from the figures for Tracks and Surfaces.

All Weather Racing. Females v Males.
Distances.
All Runners.
Distance	Sex	Runners	% Of Runners	Races	Wins.	% Of Total Wins.	A:E Figure.
5 Furlongs.	Males	8175	54.70%	1612	1007	62.47%	0.89
	Females	6780	45.30%	1612	607	37.66%	0.80
6 Furlongs.	Males	14536	57.60%	2516	1688	67.09%	0.89
	Females	10690	42.40%	2516	835	33.19%	0.77
7 Furlongs.	Males	14908	58.90%	2409	1687	70.03%	0.90
	Females	10392	41.10%	2409	726	30.14%	0.73
8 - 8.5 Furlongs.	Males	15934	59.50%	2589	1812	69.99%	0.89
	Females	10839	40.50%	2589	780	30.13%	0.73
9 - 10.5 Furlongs.	Males	9914	59.30%	1574	1083	68.80%	0.88
	Females	6820	40.70%	1574	492	32.20%	0.73
11 Furlongs +.	Males	12933	61.55%	2115	1493	70.59%	0.89
	Females	8076	38.45%	2115	628	29.69%	0.75
Performance Of SP Favourite.
Distance.	Sex	Runners	Wins	Win %	PL @ SP	ROI	P/L Lay	ROI Lay	A:E Figure
5 Furlongs.	Males	893	309	34.5%	-52.68	-5.9%	-9.58	-0.75%	0.94
	Females	448	159	35.49%	+7.17	+1.60%	-26.16	-4.00	1.00
6 Furlongs	Males	1,502	532	35.42%	-22.11	-1.47%	-75.77	-3.42%	0.98
	Females	583	188	32.25%	-53.46	-9.17%	-0.59	-0.06%	0.91
7 Furlongs	Males	1,491	511	34.27%	-84.23	-5.65%	-106.66	-5.03%	0.95
	Females	549	146	26.59%	-128.44	-23.40%	+50.00	+6.14%	0.75
8 - 8.5 Furlongs	Males	1679	625	37.22%	-21.95	-1.27%	-105.01	-4.22%	0.99
	Females	644	178	27.64%	-108.69	-16.85%	+12.35	+1.16%	0.81
9-10 Furlongs.	Males	871	278	31.92%	-81.40	-9.35%	+6.62	+0.53%	0.89
	Females	341	119	34.90%	+6.59	+1.93%	-66.24	-11.34%	1.04
11 Furlongs +	Males	1,321	472	35.73%	-61.72	-4.67%	-57.67	-3.08%	0.96
	Females	489	139	28.43%	-113.56	-23.22%	+38.15	+5.06	0.77
Summary of Distance figures.It will be interesting to follow the progress of these figures in the future. They point to Females being at a bigger disadvantage over distances greater than 6 furlongs, especially in the 7 - 8.5 furlong range, while the 5 furlong figures show a much reduced bias against Females, if any, (although there were a higher percentage of Females contesting 5 furlong races than other distances). Both sets of figures (for All Runners, and SP Favourites) seem to support each other.

All Weather Racing. Females v Males
Age Groups.
All Runners.
Age	Sex	Runners	% Of Runners	Races	Wins.	% Of Total Wins.	A:E Figure.
2yo's.	Males	12055	53.90%	2299	1479	64.33%	0.91
	Females	10276	46.10%	2300	819	35.61%	0.76
3yo's (against 3yo's only).	Males	14016	54.20%	2809	1810	64.44%	0.90
	Females	11849	45.80%	2809	1006	35.81%	0.78	*
3yo's (against all age groups).	Males	24200	52.20%	6721	3067	45.63%	0.91
	Females	22167	47.80%	6922	1818	26.26%	0.77
4yo's (in races restricted to 4yo's+).	Males	4948	49.50%	2250	593	26.36%	0.90
	Females	5041	50.50%	2397	370	15.44%	0.74
5yo's (in races restricted to 4yo's+).	Males	12793	71.38%	2714	1456	53.65%	0.90
	Females	5128	28.62%	2403	333	13.86%	0.67
* Blindly laying all 3yo fillies in mixed sex races restricted to 3yo's has shown a profit of over 100 points at level stakes (with 5% commission) in each of the past 3 years
Performance Of SP Favourite.
Age	Sex	Runners	Wins	Win %	PL @ SP	ROI	P/L Lay	ROI Lay	A:E Figure
2yo.	Males	1432	557	38.90%	-58.90	-4.11%	-16.03	-0.89%	0.96
	Females	669	238	35.58%	-53.40	-7.98%	+11.68	+1.27%	0.92
3yo's (against 3yo's only).	Males	1,776	662	37.27%	-68.48	-3.86%	-42.17	-1.81%	0.95
	Females	795	237	29.81%	-137.37	-17.28%	+35.62	+2.80	0.83
3yo's (against all age groups)	Males	930	357	38.39%	+25.92	+2.79%	-98.18	-6.33%	1.03
	Females	581	168	28.92%	-76.56	-13.18%	-11.80	-1.08%	0.86
4yo's (in races restricted to 4yo+)	Males	624	211	33.81%	+5.19	+0.83%	-79.01	-6,61%	0.99
	Females	288	103	35.76%	+23.37	+8.11%	-45.22	-8.60%	1.03
5yo's+ (in races restricted to 4yo's+)	Males	1,265	400	31.62%	-60.24	-4.76%	-68.83	-3.05%	0.94
	Females	257	53	20.62%	-77.61	-30.20%	+41.55	+7.25%	0.69
Summary of Age Groups. 2yo and 3yo Fillies racing against their own age group have the lowest figures. Both are showing a positive return when laying at level stakes.  This is consistent with the theory that females are at a physical disadvantage when racing against Males of a similar age.  There were some interesting figures for 2yo's on a year to year basis (note below). The low figures for 5yo+ Mares are harder to explain.  Because the sample size was low, I expanded the search to include every race (1-99% Males) for this age group, and checked the figures on a monthly and quarterly basis looking for any clues.  They remained constant throughout the year however, with each Quarter and each Month having similar (lay) figures. The positive backing figures at Industry SP for both Male and Female favourites as 4yo's in open company could be of interest, and possibly worth following for the future.  Females as non-favourites in these races had a poor record. I think there are clues amongst the figures for Age Groups that could shed some light on the bias against Females on the All Weather, and it is this category that interests me the most.

Year to year variance in the figures for 2yo fillies.

This category had the most erratic variance in fillies' performance from year to year.

• From 2007 to 2012 the ratio of Males to Females in all Mixed Sex 2yo Races was fairly consistent at around 58:42 (range between 59:41, and 57:43).
• There was an average of 296 Mixed Sex 2yo races per year (low 264, high 313).
• Males dominated the races won by an average ratio of 68:32 (range from 65:35 to 70:30).
• Blindly laying Fillies on a year to year basis produced the following P/L figures at level stakes:
• 2007 +38.5
• 2008 -106.8
• 2009+21.1
• 2010 +425.5
• 2011 -2.4
• 2012 +292.5

Interestingly, last year (2013) there was a sudden drop in the ratio of Males that contested 2yo Mixed Sex Races. 

The total number of Males dropped from the previous 6 year average of 1648, down to 1400, while the number of females contesting the same races stayed very close to the previous 6 year average of 1186 (at 1130). 

The ratio of Males to Females contesting these races last year was 55:45.

I haven't been able to explain this sudden drop in Males contesting Mixed Sex 2yo races on the All Weather yet, and may have to wait until the 2014 figures are available to see if this is an ongoing trend if no explanation can be found before then.

Conclusion

The figures from the tables in this blog do show a bias against Fillies and Mares when racing against Males on the All-Weather.

I've looked at each category closer than the general tables show to try and find reasons for this, but I wasn't able to pinpoint anything of significance with the limited sample sizes available.

There were no clues when looking at the time of the year for example.

In general, the figures for 5 Furlong races showed less bias than any other category, and 7-furlong races had the largest bias.

Contrary to theories that I've heard and read, there doesn't appear to be any difference in bias between Fibresand and Polytrack. This could actually give a clue as to the type of Female racehorse that competes on the All-Weather. 

Is it possible that owners and trainers simply race inferior Fillies and Mares on the All-Weather surfaces? 

The bias is not as noticeable on the Turf, and the better Females are possibly kept for that type of racing.

This theory helps to explain a number of things that have bothered me since starting this research, and it's an idea that has grown on me the more time I spend looking at the results of this research.

I'm a fairly avid follower of betting systems, and I've noticed that when eliminating all females from system qualifiers the profitability of the system improves. 

This was also suggested by Flatstats.com in an article published by them 10 years ago.

I've got plenty of figures at hand from all the research that I've done for this blog, and if anybody reading this is interested in a closer look at any particular category feel free to PM me, and I'll help if I can. 

I'd like to publish the rest of the research that I have, and will probably do so when I've decided whether or not it's relevant for value betting and finding winners.

This subject of Females against Males has captured my interest, and I feel that it's worth more ongoing study to try and determine the exact cause of the apparent bias.