All About Albot
When the ATP releases its new rankings Monday, Radu Albot is expected to be #46 in the world, a new career high, after beginning the year at #98. In Miami, he qualified, won his first round match, and then hung in there with Roger Federer for more than 2 hours, including taking the first set from him. He won a couple of matches at Indian Wells, made the semis at Montpellier, and most importantly, won Delray Beach.
There are a lot of reasons you would not know much (if anything) about Albot. He’s 29 years old, and has been hanging around the #90-#110 range for three-plus years, but before that, was deep in Challenger-land. If that’s not inconspicuous enough for you, he’s also 5’9″ (175cm) and 152 lbs (69kg), and from Moldova, a small country sandwiched between Ukraine and Romania. He’s not the only Moldovan to play on the ATP tour, but as best as I can tell he is the only one to ever rank higher than #670. If you type his last name into the search box on the ATP site, he comes up seventh, behind five Talbots and a Torralbo.
But he is a Top 50 player now, and you will be seeing a lot more of him for at least the rest of the 2019 season.
Lucky Run or a Radu-cal Change in His Game?
It’s easy — well not easy, but possible — to hit a hot streak, advance deep in a few tournaments, particularly against weak competition, and see a big bounce in ranking. Is that what Albot is doing, or has he changed his game changed in a way that helps explain how he has cut his ranking in half in about three months?
And it’s not just his ranking. As of the beginning of the year, I had Albot’s hard-court ELO at 1595. After Miami, I have him at 1781. He does not have any signature wins, but he has two over Karlovic, and one each over Kohlschreiber, Fognini, Kyrgios and Johnson, all established, quality players. Mostly, though, he’s winning a lot of matches.
2018 vs 2019 Results
In 2019 so far, he is 17-6 in ATP main draw and qualifying matches (which is the dataset I will use for the rest of this post, unless otherwise noted). In 2018, he played 28 main draw and qualifying matches, just five more than his first three months of 2019, and he was 12-15, excluding a walkover match. His results appear to permit a fairly apples-to-apples comparison, because the average rank of his opponents was about 97 last year, and this year it is about 87.
If you are skeptical about using the ATP ranks (after all, one of his losses was to #210 Tsonga), the average ELO of his opponents is consistent with the above. In 2018 his opponents had an average ELO of 1692. This year it’s 1709.
Those numbers are not what you would call a “tough schedule,” but no matter how you look at it, he’s facing slightly better competition, and getting better results. The numbers also are similar enough that we can look at some match stats from the two periods to see what, if anything, has changed in Albot’s game and contributed to his recent success.
Albot the Servebot?
Not quite. But comparing his match stats in 2018 and 2019, it doesn’t take long to notice that he’s serving better in 2019. A lot better. From the match stats alone, it appears he has decided to take more risks on first serves, and fewer risks on second serves. Take a look at this table:
That’s an amazing change in a service profile from one season to another, against similar opponents, particularly for a guy who is 5’9″.
He has vastly improved his second serve performance. He has cut his double faults by a quarter, and either improved the second serve to win more of those points, or improved what he is doing after he hits the second serve to win more of those points. I would speculate that because of the second serve performance, he is taking more chances on the first serve. His ace percentage is up almost 50%. He hasn’t made any huge gains on the first serve, because the improved 1st Won% is largely offset by getting fewer first serves in, but even a half-percent gain can matter at the top levels of the game.
When did this happen?
I would have assumed Albot worked on this during the off-season, but the data suggests it started earlier. Here’s the same table from above, except this time the first column is the January-March 2018 hard court season, and the second column combines the second-half 2018 hard court season and 2019 hard courts so far.
It looks to me like he made the sea change in his service profile between the two hard court seasons in 2018, not in the offseason between 2018 and 2019. That first column is only 14 matches, so the sample size could be better, but it represents nearly 200 service games, so should be reasonably representative of his service profile for the period (and it’s also reasonably consistent with 2017, although a little weaker in some areas).
I thought maybe after Miami 2018, which is the cutoff in the table, I would have seen that he dropped back down and played some hard court Challengers while working on the serve to prep for the second hard court season. That’s not what happened. After Miami 2018, he didn’t play another hard court match until Winston-Salem in August, which is where the second column begins.
So it appears that most of the gains he made on serve were made during his 2018 clay court (or grass) season. We don’t have as many clay matches to work with, but let’s compare his 2017 clay court season to his 2018 clay court season, to see if the change in service profile shows up there also, albeit with a different “mix” to account for the surface change. For this I have included the Challenger matches to give us a bigger sample.
It certainly isn’t obvious that he was changing his serve approach during the 2018 clay court season. His first serve is better than in the prior clay season, but his first serve percentage is up, not down. And, we don’t see anything that indicates he was working on his second serve during the clay season. It’s a radically different surface, so even if he had plans to change his serve profile on hard courts, he would not necessarily try the experiment on clay courts.
What about grass? Always a sample size problem. He only played four grass matches, and one of them was against world #881, which he lost! Looking solely at his three Wimbledon matches, he picked up five set wins against Bedene and Carreno Busta, who was then ranked #12, before losing to Isner in straights. Granted he was on grass and it is only three matches (albeit 60 service games), but his numbers there are 62.8% 1stIn / 67.9% 1stWon / 5.0% AcePct / 59.2% 2dWon / 3.7 DF% / 64.7% SPWon. With all necessary caveats in mind, it still looks like the more aggressive approach is in play by Wimbledon in July.
A Smidge of Hawkeye Data
There isn’t much granular data out there on Albot. The Match Charting Project has seven matches from this season, but none from last season (I better get on that!). There is one other place to look. As you probably know, Hawkeye data generally is not publicly-available. However, it is possible to find some piecemeal, Hawkeye-derived data on the Web. It can be error-prone in some categories, particularly in unforced error counts, distances run and RPMs on rally shots, and sometimes is missing data for certain categories. It’s also collectible only from select tournaments, and only on the courts that have Hawkeye (compare Indian Wells, where most of the courts have Hawkeye, to Miami, where only four courts have it).
It’s better than nothing.
The Hawkeye serve data I have compiled includes information about service speeds, net clearances on serves, and a handful of other miscellaneous data. I have been able to compile 13 matches-worth of that data on Albot. Unfortunately only two of those are from 2018.
The 11 matches from 2019 are from four tournaments. Here are a couple of data points from 2019 on Albot’s serve.
|1st Serve Speed||174.2 km|
|1st Serve Net Clear||0.748 m|
|2d Serve Speed||145.3 km|
|2d Serve Net Clear||0.735 m|
|Shots to Hold||16.6|
It’s too bad we do not have enough 2018 data to compare, but I will mention what I have, because it is interesting.
In the first 2018 match for which I have Hawkeye data on Albot, he’s at Indian Wells, hitting first serves at an average speed of 153.3 km and second serves at 142.8 km. Read that first number again, and compare to the immediately preceding chart.
In the second 2018 match, he’s at Beijing late in the 2018 hard court season, and he’s hitting first serves at an average speed of 179.1 km, and second serves at 143.2 km. The 2018 data does not have the net clearances for Indian Wells, so there’s no point mentioning those, and the Shots to Hold figure is particularly susceptible to variation in small sample sizes, because a lot of that depends on the quality of the opponent.
What we have shows a difference in average first serve speed of more than 25 km between Indian Wells and Beijing in 2018. The 179.1 at Beijing is higher than the 2019 data, so it may be very match-specific. Still, the more reliable 2019 data is higher than 2018 Indian Wells by 20 km, which is a big difference. The jump may just be a glitch due to a tiny sample size, Hawkeye errors or something else. We also have to be careful because everything I’ve written in this post previously has set us up to believe that something positive happened to Albot’s serve in the middle of 2018. When you see the jump from Indian Wells to Beijing, it fits the setup perfectly, but it does not prove the setup.*
*I did not intend the setup that way. I knew the 2019 Hawkeye data showed higher speeds than the 2018 average, but did not see the difference between the two 2018 Hawkeye matches (and their different points on the calendar) until I started writing this section.
The jump in first service speed does fit the more reliable match stats data that shows fewer first serves in, a higher ace percentage and a higher percentage of first serves won. But again, it’s only two 2018 matches.
There’s more, though! The Hawkeye data contains another category for “Historical Serve Speed.” We don’t have any idea what period that covers, so we have to be careful with it. It does help us a little in a situation like this, where we only have two 2018 matches. We don’t have the exact average serve speed from the other 2018 hard court matches, but we do have the historical first serve speed reported for each of the two matches in 2018 — so, the average first serve speed from some period of time prior to each of those matches as recorded in Hawkeye’s database. The historical speeds are about 160 km as of the dates of those two 2018 matches, which is some validation of the lower Indian Wells speed and a ramping up of first serve speeds later. For comparison, in the 11 Hawkeye matches I have for Albot in 2019, the historical first serve speed is reported to be around 170 km.
Although we can’t say for sure that he ramped up his first serve speed in the summer of 2018, the two Hawkeye matches we have from 2018, and the historical averages that Hawkeye records, suggest something significant was happening with his approach to the first serve, and that is supported by the match stats.
What about the Second Serve?
The Hawkeye data does not give us anything meaningful to compare for Albot on the second serve. The speed of the second serve may not be particularly important anyway. The net clearance data would be nice to have from 2018, as more net clearance in 2019 could support the significant reduction in double faults, perhaps because of more topspin or kick on the second serve. But we don’t know.
What we do know is that Albot has vastly improved his second serve results. Not only that, but his second serve results are well above Tour averages at this point, and are knocking on the door of elite company.
With his ranking jump, we will be seeing a lot more of Albot in tournaments because he has direct entry. And if he keeps serving like he has been in the last 7 months, we will see him deeper in tournaments too.