Which picks winners better?

Deciphering Winning Strategies: Humans vs. Algorithms & Data Driven Tools in Horse Racing

Hi guys, in the realm of horse racing, a profound debate unfolds: Are humans or algorithms better at identifying the winning edge? Can a machine really outsmart the sharpest human minds in racing? Or is there something about experience, intuition and gut feel that no amount of code can replicate? This debate is reshaping how winners are found.

Here I explore and delve in to the intricacies of both approaches, shedding light on their respective strengths and capabilities. Data-driven tools and algorithms possess a remarkable ability to unveil complex patterns and correlations hidden within vast datasets. They excel in multivariate analyses, calculating probabilities, and providing objective comparisons across races and horses. With consistency and speed, algorithms deliver real-time updates and highly objective, though not always complete, decision-making and instantaneous calculations. These insights are invaluable, revealing trends and anomalies that might elude human perception.

Data-driven tools in the context of horse racing refer to software, algorithms, and analytical techniques that utilise vast amounts of data to extract insights, patterns, and correlations. These tools leverage computational power to process and analyse data in ways that go beyond human capabilities.

Here are some aspects of data-driven tools used in horse racing:

Pattern Recognition: Data-driven tools can identify complex patterns and trends in large datasets that might not be immediately apparent to humans. They can analyse historical race results, horse performances, jockey-trainer combinations, track conditions, and more to identify consistent patterns of success or under-performance.

Statistical Analysis: Algorithms can perform sophisticated statistical analyses to quantify relationships between different variables. They can calculate probabilities, odds, and expected outcomes based on historical data and current conditions.

Machine Learning: Machine learning algorithms can learn from historical data to make predictions about future races. These algorithms can adapt and improve their predictions over time as they encounter more data and learn from their mistakes.

Data Integration: Data-driven tools can integrate data from various sources, such as race results, horse pedigrees, jockey and trainer statistics, track conditions, and even external factors like weather and news. This comprehensive view allows for more informed decision-making.

Real-Time Analysis: Some data-driven tools can provide real-time analysis of on-going races. They can consider factors such as live odds, in-race performances, and other dynamic information to adjust predictions and provide insights as the race unfolds. The Betfair Exchange platform, for example, offers live betting markets and real-time odds fluctuations during races. Users can monitor the market movements and adjust their bets based on the changing odds and in-play performances.

Anomaly Detection: Algorithms can identify anomalies or outliers in the data that might indicate unique situations or potential value bets. For instance, if a horse with consistently high performance suddenly has odds that seem undervalued, the algorithm might flag it as an anomaly worth investigating. Proform Racing, for example, offers a range of analytical tools, including anomaly detection features. Their software can analyse historical data and identify anomalies in horse performance, odds movements, or other relevant factors. Users can set up custom alerts to be notified of potential value bets or unique situations in real-time. Proform Racing is geared towards serious bettors and professional handicappers/tipsters/pundits that are looking for advanced analytical tools to gain an edge in their betting strategies. It caters to users who want to delve deep into horse racing data and develop their own betting systems.

Risk Management: Data-driven tools can help assess the risk associated with different bets. By analysing historical outcomes and potential scenarios, these tools can assist bettors in making more informed decisions while managing their risk exposure. Geeks Toy is a popular trading software package used by many professional punters on Betfair Exchange. It offers advanced features such as tick offsetting, which allows users to automatically place opposing bets to lock in profits or minimise losses as market conditions change.

Back-testing: Algorithms can be used to back-test strategies against historical data. This allows analysts and punters to see how a specific betting strategy would have performed in the past, helping to refine and optimise their approach. So, you develop your own strategy or system using something like Proform Racing then back-testing allows you to refine it by simulating how those strategies would have performed in the past. By analysing historical outcomes, you can identify patterns, trends, and potential inefficiencies in the market that your strategy could exploit. You can then optimise various parameters, such as entry and exit criteria, staking methods, or selection criteria. By testing different combinations of these parameters against historical data you can identify the most effective settings for maximising profitability. Back-testing also serves as a validation tool to assess the robustness and reliability of a betting strategy and allows you to evaluate the risks associated with it.

It’s important to note that while data-driven tools offer powerful insights, they are not infallible. They require well-defined parameters, accurate data, and occasional human oversight to interpret the results and adjust for unique situations that might not be captured in the data. Harmony between human expertise and data-driven analysis ultimately leads to more well-rounded and informed decisions in the field of horse racing.

Here are some examples where algorithms excel when it comes to horse racing:

Track Bias and Conditions Impact: Algorithms can analyse the performance of horses on different tracks and under various conditions (such as track surface, weather, distance) to identify subtle biases or preferences. For instance, an algorithm might discover that a particular horse consistently performs better on rain-softened turf tracks than on firm ones, which might not be immediately evident by merely looking at individual race results.

Jockey/Trainer Combinations: Algorithms can examine the historical data of jockey/trainer combinations and identify specific pairings that consistently lead to improved performance. This might involve uncovering a trainer’s tendency to excel with certain jockeys on specific types of horses, which might not be apparent without extensive data analysis. I regularly factor such stats in where you might see me say in a write-up something like, “Aidan O’Brien & Ryan Moore have a 60% strike rate at the track.”

Stamina Trends: Algorithms can analyse a horse’s past performances to identify stamina trends over different distances. It might discover that certain horses tend to perform better when they gradually step up in distance, revealing insights into the horse’s potential for longer races. I applied this trend to good effect at Cheltenham where I considered, along with the knowledge that Gold Cup winners really need to be able to stay that 3m2f extremely well, the stamina of Grand National winner Corach Rambler before tipping him at 22/1 ante post.

Pace & Running Style Analysis: Algorithms can dissect the race dynamics and the running styles of horses to uncover patterns related to their positioning throughout a race. For instance, an algorithm might reveal that horses with a front-running style tend to perform better on specific tracks with short run-ins, where tactical positioning is crucial. I tend to use information like this at tracks like Newcastle where the Tapeta surface there can be favourable to front-runners, particularly if they can establish a strong pace and handle the track’s long straight, where it can be difficult for horses to make up ground from behind. Also at Chester, where draw bias is a significant factor with those drawn low particularly in 5f-6f races have a major advantage, and the tight turns with the undulating nature of the course often make it difficult for horses to come from behind, giving an advantage to front-runners who can maintain a strong pace and position.

Form & Layoff Patterns: Algorithms can detect patterns related to a horse’s form after a layoff or extended break from racing. It might identify horses that consistently perform well in their first race back from a break, indicating the ability to maintain fitness during downtime. Humans can do this too but it takes us much longer.

Stall Draw Influence: Algorithms can assess the impact of starting gate positions (the stall drawn by each horse) on a horse’s performance in different races. It might reveal that certain tracks or distances favour horses starting from specific draws due to track layout or surface conditions. Chester, as already mentioned, is known for its significant draw bias, especially in races over shorter distances while Goodwood’s undulating track can favour horses drawn high, particularly in longer distance races such as the 1-mile contests.

Race Class Transitions: Algorithms can identify horses that excel when moving up or down in class. They might uncover instances where a horse performs better when moving from claiming races to allowance races, suggesting adaptability to changing competition levels.

Weight Effects: Algorithms can analyse the impact of varying weights carried by horses in different races. They might reveal horses that consistently outperform their peers when carrying higher or lower weights, shedding light on the importance of handicapping factors. I know from experience this is a particularly difficult/time consuming factor for humans to spot/assess.

These examples showcase how algorithms can uncover intricate relationships between variables, spotting trends and anomalies that might be hidden within the vast amount of data. By recognising these patterns, algorithms assist humans in making more informed decisions in horse racing analysis and betting strategies.

On the other hand, experienced human analysts possess a unique depth of insight that algorithms cannot replicate. They are skilled at identifying unquantifiable factors like a jockey’s mind-set or the impact of weather and track conditions on specific contenders. Human intuition extends to observing race horse body language, understanding jockey intentions (particularly useful just before the start of a jumps race, for example, as you can see by where the jockey is positioning his horse as they line-up whether he intends to front-run/be positive on the horse or not), and integrating recent news and events that could sway the odds. Another example, a trainer’s bullishness in an interview about the chances of his runner that won’t show up in raw data. A human’s ability to grasp in-depth context, recognise unusual circumstances, and compensating for the impact of personal/cognitive biases lends a dynamic layer to the analysis that algorithms might lack.

This is also where I personally add a layer of value each day by checking Oddschecker.com, which lets me quickly compare bookmaker prices, find standout early odds, and identify which firms are offering extra places or promotions. It’s a small edge that, combined with experience and timing, can make a big difference to long-term profitability.

So, what is the solution?

The Smartest Approach: Combine Both

It’s not a question of humans or algorithms.
The best punters and tipsters already know it’s humans and algorithms.

Algorithms can feed you patterns, probabilities, and red-flag warnings.
But humans can interpret those signals with context, wisdom, and experience. This synergy allows for identifying true value and mitigating risks that neither approach could achieve alone.

You want to leverage the strengths of both. When the precision of data-driven analysis combines with the nuances perceived by human intuition, a comprehensive approach emerges. The synergy between human expertise and algorithmic insights leads to well-rounded, better informed decisions. By integrating the data-rich outputs of algorithms with the seasoned wisdom of human analysts, the path to picking winners becomes more navigable.

The Racing Post, a prominent horse racing publication, for example, uses a combination of human expertise and algorithms to help choose their selections or tips that they publish. They employ a team of experienced racing analysts who study the form, performance history, and other relevant factors of horses and races. These analysts utilise their knowledge and expertise to provide insights and predictions for races. However, the Racing Post also employs algorithms and data-driven tools to enhance their analysis and selections.

By combining the expertise of human analysts with the computational power of algorithms, the Racing Post aims to provide more accurate and comprehensive selections and tips for their readers. This approach leverages the strengths of both human intuition and data-driven analysis to offer valuable insights to those interested in betting on horse racing.

This is also what I try to do everyday — using powerful data tools behind the scenes but adding the extra human judgement on top to find, filter, and finesse value selections for you.

Final Thoughts

Algorithms bring incredible power to the table — but they aren’t magic on their own.
Human intuition brings depth and nuance — but alone it can miss hidden patterns.

Together?
In the ever-evolving landscape of horse racing, the partnership between human and algorithm continues to shape the future of successful betting strategies.

If you enjoyed reading this or found it insightful any “likes” or positive feedback in the comments section below would mean a lot to me. 🙏

Jibber Jabber

If you like my write-ups and they’ve helped you then please…