Friday, 10 July 2015

Review of Roger Smith’s ‘Analysing Formula 1’ (2008)

DATE: January 2008

This book is an enjoyable read, for Roger Smith is obviously a keen and knowledgeable enthusiast who has delved into racing. Smith claims not to have based his findings on simple wins or win rates and purports to consider the effects of the car and the opposition – factors crucial to gauging driver performance. For someone who is not a journalist, Smith’s text is remarkably proficient. The book bears some uncanny similarities to my analytical book, Grand Prix: A Century of Racing (published by AA Racing, November 1994), in terms of being based on descriptive statistics in the form of simple bar graphs, win rates, rivals wins, grid position vs wins, etc.

Despite the impressive graphs, matrices and explanations, Smith’s result proclaiming that three drivers ranked at the top – Fangio (career 1950-58), Clark (career 1960-68) and Schumacher (career 1991-2012) – makes one question the methodology. Common sense alone tells us that it is not possible to directly compare drivers across eras, let alone actually rank them. How can it be stated with any certainty that Fangio was better than Schumacher and Clark? Or that these three are better than any of Alonso, Senna, Prost, Lauda, Andretti, Piquet, Stewart, Surtees, Moss or Ascari? There are too many incomparable variables in each career that defy comparative measurement.

In contrast, my Rating System ranks drivers in tiers based on my analysis of their comparative time-based speeds. By my System the fastest drivers of each era rate equal in the first tier, from Fangio, Ascari and Moss through to Prost, Schumacher and Alonso. I certainly would not be able to rank them due to incomparable variables. It is not possible to rank any of the top drivers across eras as greater or as the greatest. They never faced each other and cannot be directly compared.

Smith claims to have taken the quality of the car into account (“it is rare that a great driver can prosper without a great car, or vice versa” (p. 10)), yet some of his findings seem to be results-based data, bolstering the dominance of the drivers who he deemed to be in the ‘top three’, yet he neglects the fact that these three drivers, Fangio, Clark and Schumacher, had superior cars for longer periods of time than most.

When Fangio, Clark and Schumacher did not drive the best cars, their results were not dominant: Fangio-Maserati in 1953, Clark-Lotus in 1961 and 1966 and Schumacher-Ferrari 1996-99 and 2005. Proof that no matter how talented a driver, his performance is largely dictated by the competitiveness of his car.

The many variables in F1 racing make it so complex that any comparative analysis has to be detailed and a methodology consistently applied. Not only does the car largely determine a driver’s competitiveness, but each car model’s performance varies on different circuits, and when upgraded and modified, within and across seasons. Such a complex set of variables significantly impacts results. On this score, Smith appears to have made some generalised, sweeping assumptions to inform his analysis.

For instance, the Maserati 250F seems to have been treated as one model, Smith stating that in 1957 Fangio drove the 250F Maserati “entering its fourth season and nearing the end of its competitive life”. The 250F was in fact made in five quite differing versions each with varying levels of performance relative to the front-runners:

● the initial, belouvred 1954 model
● the upgraded, unlouvred and more competitive 1955 version with which Moss won the 1956 Monaco GP
● the ‘low-drag’ 1956 model influenced by the Vanwall’s aerodynamic shape
● the ‘offset’ 1956 model with which Moss won the GP at Monza
● the final 1957 version which Fangio drove to his fifth crown, which was by far the most competitive of the 250Fs in relation to the opposition.

My Rating studies compare on-track time differentials only, and do not rely on wins, results, places, etc. For example, I discovered that the performance between the various Maserati 250F models varied in cases by over 1.0%. Considering that the winning margins (seasonal averages) for championship-winning packages have been 0.2%–0.5%, incorporation of such detailed factors is crucial to any comparative analysis. Fangio’s debut win in the 1954 Argentine GP with the first 250F was unfairly achieved. The Frana and Gonzalez Ferraris had been leading and faster almost throughout,. When Fangio pitted, more than the permitted number of mechanics attended him. Ferrari protested and slowed their two cars. Fangio swept past and his win was wildly applauded and the officials did nothing about Ferrari’s legitimate protest. Yet this win would have contributed to Fangio’s tally… In 1957 Fangio was driving the fastest car of the season, and the most competitive of all 250Fs; it most certainly was not ‘nearing the end of its competitive life’ as Smith asserts. Fangio enjoyed a wide margin of car superiority over the rival Vanwall and Lancia-Ferrari cars in 1957. Bare race results do not distinguish such qualitative ‘quirks’ of racing performance behaviour.

Similarly, Smith treated the Lotus-Cosworth 49 as one model. It was in fact built in three basic versions of widely varying perfromance in relation to its rivals:

● the original Type 49 of 1967 and driven to Clark’s last win at Kyalami in 1968;
● the longer wheelbase 49B of 1968, later fitted with aerofoils, which Graham Hill drove to his second title; my calculations showed this car to have been a whopping 1.5% faster than the first short-wheelbase version.
● the final 49C driven until early 1970 by Rindt.

There is no indication in Smith’s book that the widely differing performance of these Lotus 49 variations has been factored in.

One significant model variation clearly ignored by Smith is the 1989 McLaren-Honda MP4/5. Smith states of Prost and Senna: “over those two seasons 1988 and 1989, they started 32 races together in identical machinery”. This was only true of 1988; for 1989 Senna was provided with superior engines, such that, from parity race performance in 1988, Prost was bizarrely 0.3% slower in 1989 and in qualifying suddenly a whole 1.0% slower than in 1988. Only Prost’s car set-up skills and sensitive, intelligent racing genius reduced the huge 1.0% qualifying gap to 0.3% for the races. At the time, nobody in the McLaren-Honda team knew about this engine favouritism – except of course Honda who did not divulge that they were favouring Senna. Prost stated: “No way is a driver suddenly a second a lap slower”. Prost’s complaints about Senna having more power were dismissed by the British press as ‘whingeing’. Years later, one of Honda’s engineers confessed that Honda had indeed favoured Senna over Prost. Smith has not differentiated the data for this 1989 season, which invalidates half of the information used in the Senna-Prost comparison, during the only two years in which they were team-mates.

The varying performances of models across seasons can significantly affect driver results. For example, when the dominant 1978 Lotus-Cosworth 79 competed in 1979, its performance relative to the front-runners was way below what it had been in 1978. Similarly with the 312T3 Ferrari in 1979, the Ferrari F2004 in 2005 and many other existing models when carried over into a new season. Smith’s work glosses over such distinctions, jeopardising the validity of his measurement methodology and his findings.

Another generalised statement that casts doubt on Smith’s data informing his rankings is the following: “Moss was Fangio’s equal at best”. Yet when Moss and Fangio raced the Maserati 250F/57 as team-mates at Buenos Aires in January 1957, Moss outqualified Fangio by 1.1 seconds! In the race and after an unscheduled nine-lap pitstop for repairs, Moss made up two laps on “The Old Man” (Juan Fangio). This was not indicative of equals! This highlights another variable that needs to be taken into account: driver competitiveness is not consistent over a career, even for the best drivers. Usually rookies build up to their peak after three to five seasons, and older men’s speed tapers towards career end. This was why Moss was faster than Fangio by 1957. As Niki Lauda said of his own last two seasons 1984-1985: “You cannot go at it as you did when young” and Prost of his 1988-1989 seasons: “Ayrton [Senna] was faster than me; after ten years you no longer go flat-out” (quotes paraphrased from magazine articles). Much depends on which stage of their respective careers drivers faced each other. This factor also precludes direct driver comparisons.

Roger Smith correctly states: “Senna’s rivalry did for Prost’s aspirations just as Prost did for Senna” – that is, they spurred each other on. Yet the crucial factor of competitive team-mates does not seem to have been taken into account with the three whom he finds superior to all: Fangio, Clark and Schumacher. Uniquely all three were spared competitive team-mates for most seasons throughout their careers. This means that their win rates and other statistics would be superior to those of drivers who had tougher team-mate competition, such as Senna and Prost, Piquet and Mansell, Hamilton and Rosberg or Moss and Brooks.

Fangio’s most competitive team-mates were Farina in 1950, Gonzalez in 1953 and Moss in 1955. However, at age 44, Farina was too old to really threaten Fangio, being 0.6% slower on season-average qualifying times, while the two younger men – Gonzalez and Moss – had too much deference for Fangio, who was almost a generation older, to really challenge him. In sum, Fangio was never seriously challenged in-house. Similarly with Clark; apart from 1960 when Surtees drove part-time for Lotus, Clark’s Lotus team-mates were notoriously way off-pace, not only as drivers but in equipment; the Number Two Lotus cars during Clark’s heyday were neglected by Colin Chapman (owner of Lotus and brilliant designer) and consequently so much slower and drastically unreliable. Lotus was really a one-man team. Similarly Schumacher was unchallenged, having self-selected his team-mates throughout his career, making sure they were not too competitive a threat.

There is no doubt this lack of in-house team-mate threat inflated Fangio’s, Clark’s and Schumacher’s results. In this respect their careers were so unlike those strong pairings that diluted each driver’s results, such as: Senna and Prost in 1988-1989, Moss-Brooks 1957-1958, Fittipladi-Peterson 1973, Scheckter-Villeneuve 1979, Prost-Lauda 1984, Mansell-Piquet 1986-1987 and Alonso-Hamilton 2007. How then can one compare the Fangio-Clark-Schumacher ‘sheltered’ career results with those of the other great drivers?

At least as important as a competitive, unhobbled team-mate are competitive rivals. Fangio, Clark and Schumacher were unable to dominate when they faced other top-rate drivers in good cars.

Fangio was fortunate during part of his career that his main rival and driver-equal Ascari was neutralised:

● in 1950 by Ascari’s non-competitive Ferraris in 125C2 and 275 forms;
● in contrast was 1951 when the Ascari/ Ferrari package was almost as fast as the Fangio/Alfa Romeo and scored two wins to Fangio’s three.
● In 1953 Fangio’s Maserati was slower than Ascari’s Ferrari and Fangio managed just one, perhaps fortuitous, win at Monza.
● In 1954 Ascari missed most of the season, his Lancia only appearing for the final race at Barcelona, where he out-qualified Fangio’s Mercedes-Benz by a whopping 1.0 second and led until clutch failure. Earlier in the 1954 season Ascari had twice been released for one-off drives, in a Maserati 250F for the British GP and a Ferrari 625 in the Italian GP at Monza, which latter he led for 49 of the 80 laps in a titanic struggle with Fangio’s Mercedes-Benz. On both occasions Ascari’s cars failed to finish.
 Ascari crashed fatally after the first two races of 1955, thereby removing one of Fangio’s two greatest rivals.
● Fangio’s good fortune extended to the 1956 and 1957 seasons when his then strongest driver rival Moss was hampered by his slower Maserati and Vanwall cars.

These variables do not detract from Fangio’s undoubted ability, but they certainly boosted his results and negate valid and direct driver comparison with Ascari and Moss.

Clark was similarly smiled upon by good fortune and fate. From 1960 to at least 1968, his main driver rival and equal was in fact John Surtees. Barely acknowledged in Smith’s book, Surtees was the most hampered-by-poor-machinery great driver of all. Except for his initial four rookie appearances in a Lotus-Climax 18 in 1960, when he managed one second place and was leading Moss’ similar car before crashing on slippery tramlines in Oporto, and the first two events of 1966 with the Ferrari 312/66, Surtees drove inferior cars for the rest of his 13-year career! Consequently, Clark drove in a ‘partial vacuum’ and his results cannot be validly compared with Surtees’.

Michael Schumacher enjoyed the longest competitive career of any Formula driver to date, until he retired in 2006 – a span of 16 years. For the first three seasons, Schumacher did not have the fastest car, but still scored a few wins. In early 1994, Schumacher’s Benetton-Cosworth package was virtually equal in speed to Senna’s Williams-Renault package; after Senna’s death, and for the remainder of the 1994 season, the Schumacher/ Benetton-Cosworth was 0.5% clear of Damon Hill’s Williams-Renault package. For 1995 this advantage was reduced to a still comfortable 0.3% over Hill’s Williams-Renault package. No surprise then that Schumacher won titles in both years. Schumacher’s initial Ferrari season, 1996, saw his package 0.3% slower than Hill’s Williams-Renault; from 1997 to 1999 he was on par with Villeneuve’s Williams-Renault and then Hakkinen’s McLaren-Mercedes. He was not dominant in these four seasons. Just as with Fangio, Ascari, Clark, Andretti, Stewart, Lauda, Prost and Senna, Michael Schumacher only dominated when he had the best car. As with any driver in an off-pace car, Schumacher suffered in 2005, managing a farcical, single win at Indianapolis. For 2006, Schumacher and his Ferrari were 0.1% down on Alonso’s Renault, but as always Schumacher drove brilliantly to even-score with seven wins. Schumacher’s career reads very much like Clark’s and Fangio’s in terms of package competitiveness; all three enjoyed several seasons of car-team dominance. Which negates direct comparison with others who did not.

Roger Smith claims to have measured or quantified driver results based on the three elements that affect driver results: car, team-mate and opposition. However, Smith’s statistics-based data would inevitably favour his top three drivers – Fangio, Clark and Schumacher, who score higher than any of the other top drivers. How can one say Fangio is better than Stewart or Prost, for example? They never raced each other therefore they cannot be compared. Fangio’s 1954-55 team-mate Karl Kling told how Moss was much better at setting up the car, that Fangio was not interested; how would this benefit Stewart, Lauda and Prost who were famed for their knowledge of the technical side of F1? Would Fangio’s natural talent have been enough against these three? There are no absolute answers but certainly doubts over claiming who was greater.

Below, I set out some of my own statistics under the titles Smith mentions as relevant to driver assessment:

1. Percentage of career without top-rate team-mate: Fangio and Schumacher 100%, Piquet 92%, Clark 90% , Ascari 83%, Moss and Senna 82%, Stewart 78%, Prost 46% and Lauda 38%.

2. Percentage of career without top-rate rival package: Fangio 29%, Ascari, Clark, Schumacher 25%, Stewart 22%, Senna 18%, Lauda, Prost 8%, Moss and Piquet 0%.

3. Percentage of career driving multi-winning car (my definition being a minimum total of 3 wins per season by all team drivers): Fangio 85%, Senna 80%, Prost 78%, Schumacher and Stewart 75%, Clark 62%, Ascari 60%, Lauda 53%, Moss 45%, Piquet 30%.

Any results- or statistics-based method would place Fangio, Clark and Schumacher at the top, due to the three major factors that Smith mentions: car, team-mate and opposition. In contrast, Senna-Prost 1988-1989, Mansell-Piquet 1986-1987, Scheckter-Villeneuve 1979, Fittipaldi-Peterson 1973, Hamilton-Button in 2012, Hamilton- Rosberg 2013, were close-matched team-mates and therefore took wins off each other i.e., they ‘cancelled out each other’s results. One could say that Senna and Prost were racing against Schumacher in 1990-1994, for example, therefore it *is* possible to compare Schumacher with Senna and Prost. However, they were in different cars (Schumacher in a Benetton-Cosworth up until 1995 and thereafter in Ferrari; Senna in the McLaren-Honda and Prost in Ferrari and Williams-Renault). The incomparable cars meant that Schumacher was not directly comparable to Senna and Prost. In contrast, my Rating System separates the performance of the driver from the performance of the car, in order to compare drivers directly. 

In my opinion, Smith’s analysis is fundamentally results-based, proven by his ranking Fangio, Clark and Schumacher at the top. The three top drivers who never had an equal team-mate, who had superior cars for the longest period, and who enjoyed a greater percentage of their races without strong rivals were: Fangio, Clark and Schumacher!

In contrast, my Rating System identifies those drivers who were the fast but whose results were stunted through not having a fast enough car – prime examples being John Surtees (career 1960-1972) and Stirling Moss (career 1951-1961).

In addition, the incomparability of careers that are individually and intrinsically so disparate, make direct driver rankings – such as those Smith has attempted – impossible.

This very interesting and potentially informative topic has been treated in a surprisingly loose and incoherent manner for a professional statistician such as Roger Smith. Smith computes the number of winners from “the front row”! The front row in early times (1950s-early 1970s) consisted of three, four or five cars, whereafter it reduced to three and then to two cars. For the past 40 years or so, the front row has really been one car on the so-called staggered grids. Consequently Smith’s grid time intervals would show wide variations .He therefore compares apples and oranges, thereby nullifying comparative analysis. To be more accurate, the grid should have been segmented based on time- or percentage-intervals. By contrast, the method used by Smith is spurious.

Would Schumacher’s renowned team-involvement skills and other out-of-cockpit management overcome Fangio or Clark, or would their respective natural talents have prevailed? What of those other acknowledged car set-up and technical development specialists, Surtees, Stewart, Andretti, Lauda and Prost: how would they fare against Smith’s top three in equal equipment? What of the phenomenally fast and successful Ascari, Moss and Senna? No conclusive answers are possible.

The effect of the car on package performance is far more than that of the driver, unpopular though this fact is in today’s media-driven cult of the hero and the celebrity. Nico Rosberg correctly summed up the perennial and unequal situation in F1 racing (Cape Times, 20 March 2008): “...there are six or seven cars in front that we can’t beat […] That’s just the way it is in F1. It sucks, but that’s the way it is.” As shown above, not even Fangio, Ascari, Moss, Surtees, Clark, Stewart, Prost, Senna or Schumacher can do much about car inferiority when facing other top-rated drivers in good cars.

Unpalatably to many, race results – such as Smith appears to use – have always had far more to do with the lucky timing of driver-car combinations and circumstances than with the popularly perceived but seldom occurring direct, level-playing-field, head-to-head and measurable competition between drivers.

It is therefore not possible to differentiate or to make fine comparisons between ‘Top Drivers’ with any degree of quantifiable and convincing certainty. Smith’s work is definitely a step up in analysis from the popular ‘driver-ratings’ which are characterised by being lineally ranked and scientifically woolly – being grounded in nothing more than emotion and personal opinion. However, Smith’s ultimate conclusions – that Fangio, Clark and Schumacher were the top three Formula One drivers – remain an illogical and unsubstantiated leap.

© Patrick O’Brien. Nothing from this page can be used without the permission of Patrick E. O’Brien.