I am using the 3B/2B ratio as a proxy for speed (more likely both speed and base running ability). This is probably better than just looking at triples because a guy can be fast and if he is not a good hitter or does not it the ball very hard (like Otis Nixon), he won't get many triples. But he probably won't get many doubles either. So if a player hits a ball that can get him at least a double, maybe a triple, the faster players will make it to third and the slower players may not even try. If a player gets thrown out at third base trying to stretch a double into a triple, he gets credit for a double. Most likely it is the slower runners who get thrown out trying for the triple. Whatever the case, on balls with the potential for extra base hits, the faster players will rack up more triples and fewer doubles (I looked into other ratios like the double-to-single ratio, but it did not seem to tell me anything).

I first took all players in history with 5,000 or more plate appearances (PAs) using the Lee Sinins Complete Baseball Encyclopedia (for this one it included pre-1900 players). I found their triples relative to the league average and then their doubles relative to the league average. For example, Lance Johnson had a relative triple rate of 387. That meant he hit 3.87 triples for each one an average player hit. He had a relative double rate of 68, meaning he hit only .68 doubles for every one double the average player hit. So his overall relative triple-to-double rate was 5.69. The table below shows the best and worst 25 all-time.

One thing that is strange is that most of the players whether the best or the worst played in recent times. It seems strange that the relative fastest and slowest players pretty much played recently. It might have something to do with how the rate of triples and doubles has changed over time.

The graph below shows the number of triples per PA over time. Triples have been in a long decline since 1900 or so. The rate seems to have flattened out in the last 10 years.

I also looked at triples as a percentage of all non-HR hits and the 3B/2B ratio over time. Their graphs look similar to this one. Doubles, on the other hand, have been on the rise. The graph below shows doubles per PA over time.

Doubles have been on the rise for some time and are practically to the all-time high.

What might account for so many of best and worst coming from recent times is that with a low triple rate, it is easy to go way above it percentage wise (and be way below average in doubles). So we might be likely to find guys like Lance Johnson who have a high triple rate relative to average and vice-versa on doubles. Then their relative 3B/2B number is very high. It seems like it would be the opposite case with the slow guys. With a low triple rate these days, it would be hard to go far below it. It might also be hard to go above the double rate. But when some rate like doubles is high, the spread can be high, so some players have extremely high double rates. Then if they don't hit many triples, they look slow. I am not sure about all this so it is just speculation.

In an attempt to avoid these potential problems, I used standard deviations (SDs) instead of relative scores. SDs give a good, probably truer, measure of the statistical spread of the numbers. So I found all the league averages and SDs for both doubles and triples (both per AB) for all batters with at least 100 at-bats (ABs) in each season. Then I cut that down to all the players who had at least 4,500 PAs (ABs + walks). In this case, both the 2B rate and the 3B rate were per AB.

For each player in each season, I found how many SDs above or below the league average or mean they were in both their double rate and triple rate. For example, if a player had a 10% double rate when the league average was 5% and the SD was 2.5%, he got a 2 for doubles for that year (like a Z-score of 2). But this was done for triples as well and the two Z-scores had to be combined to get a number that measured speed. To get that, I subtracted the double Z-score from the triple Z-score. For example, if a player had a triple Z-score of 1 and a double Z-score of -1, then his speed score was 2 (since 1 - (-1) = 2). That would be a fast player since this indicates that he hit alot of triples relative to the league average but not many doubles.

Then I found each player's career speed score. One more adjustment was made to get an adjusted speed score (ADJ SC). I took into account the side of the plate each batter batted on. Lefties should have an easier time making triples. Righties and switch hitters were adjusted upwards. Since lefties had an average speed score of .017 and righties had -.2347, I added .2517 (.017 + .2347) to the score of each righty. I also added .0839 to the score of switch hitters since about 67% of pitchers are normally lefties. Since the switch hitters would only bat about righty about 33% of the time, they got one-third the adjustment the righties got (actually, the average score for switch hitters was .32 but there seems to be no reason why they have an advantage and there were only about 90 of them, while there 856 players in total).

The table below shows the leaders. The B column indicates how they batted.

There is a better mix of best and worst time wise than in the first ranking. One strange thing is that 40% of the top 25 are switch hitters while they only made up about 11% of the overall group. Some older players who make the top 25 now are Combs, Critz and Carl Reynolds. There were 856 players. The complete rankings are posted nearby as a diary entry called "Complete Speed Rankings."

One think I did not take into account was park effects (we don't have park effects for both 2Bs and 3Bs going back to 1900). Roberto Clemente ranks very high (34th) and most of his career was played in Forbes Field, which had a huge outfield and I think some odd angles. It was a haven for triples. Pie Traynor (1.34, 40th) and Lloyd Waner (1.36, 39th) also were Pirates (teammate brother Paul, only had .20, 340th). Other players might have been hurt by small parks. Jackie Robinson had an ADJ SC of -.326. We have an image of him being a fast, daring and cunning base runner. But Ebbets field was maybe smaller than average parks, so hitting triples might have been harder. His career 3B/2B ratio was .198 while the average was .212. Lance Johnson had a 3B/2B ratio of .849 in home games and .54 in road games, while the league average was .116.

Clemente had 96 triples in home games and only 52 in road games. Clemente's 3B/2B ratio was .275 in road games from 1957-69 (the years available in Retrosheet). The league average was .204 during those years. Willie Wilson had 88 triples in home games during his career and 59 in road games. His career 3B/2B ratio was .404 in road games while the league average was just .139.

For Clemente, just using road games, his ADJ SC was .63 from 1957-69. This is a good score, but far below his career total of 1.42. For Willie Wilson, his ADJ SC in years with the Royals was 2.50. That would still rank very high all-time.

Some surprises in the bottom 25 are Tris Speaker and Nap Lajoie. Speaker had over 200 triples and over 400 SBs. Lajoie had 163 triples and over 300 SBs. They would not be the first players to come to mind when listing the slowest players ever. Yet here are their 3B/2B ratios with the corresponding league averages in parentheses

Speaker .280 (.339)

Lajoie .248 (.408)

Why would these guys be so far below the league average? Is there more to it, like base running instincts or knowledge? Speaker split most of his time between two teams, the Indians and Red Sox. From 1909-15 (with Boston), his 3B/2B ratio was .435 while the average was .427. From 1916-26 (with Cleveland), those numbers were .222 and .303. Maybe triples were tough to come by in Cleveland. But he was just average in Boston. Lajoie played the bulk of his post 1900 career in Cleveland.

A cursory look at the numbers indicates that from 1901-26, the Indians overall seemed to be about average in triples. Some years were above, some below. The last 2-3 years seemed bad for them for triples. In 1926 they had just 49 triples while the league average was 71. It is possible that the Indians really had good triple hitters all those years but were held down by their park. But over a 26-year period, that seems unlikely.

Ty Cobb also did not do well. His ADJ SC was just -.0476, meaning average. His career 3B/2B ratio was .408 while the league average was .340. That is pretty good. But in normalizing with standard deviations, he falls to about average. Maybe park effects hurt Cobb, but I really don't know. Like Jackie Robinson, Cobb is known for being a daring and cunning base runner. Maybe their daring running turned some singles into doubles, but if they were too daring, they might have tried to many times to turn doubles into triples and got thrown out. All of that would lower their 3B/2B ratio. But there does not seem to be any reason why they would not get thrown out too much at second base going for a double but get thrown out a lot at third trying to turn doubles into triples.

If anyone wants the complete rankings of the 856 players, go to

Additional sources: Retrosheet and the Sean Lahman Database

## Connect with Beyond the Box Score