SWOLF: What It Measures, and What It Rewards

Seconds plus strokes: a sum that should not be legal

The definition is as simple as a metric gets. Swim one length, note the time in seconds, count the strokes, add them together. Thirty seconds and twenty strokes gives 50. The name is a portmanteau of swim and golf, and the golf part is the useful half of the analogy: it is a score, lower is better, and it has no units because it is not really measuring a physical quantity at all.

It is worth saying plainly why that last clause is true, because most explanations skate over it. Seconds are a duration; strokes are a count. Adding them is like adding your age to your shoe size — the result is a number, but it is not a measurement of anything, and there is no principled reason the exchange rate should be one stroke to one second rather than one to two or one to a half. SWOLF is an index, not a quantity. Every property it has, good and bad, follows from that.

So why does it work at all? Because the two terms are not independent. Swim more efficiently and you cover the length in fewer strokes and less time; swim worse and both go up. When both terms move the same way, their sum tracks the underlying quality faithfully, and the arbitrary exchange rate never comes up. The whole risk sits in the cases where the two move in opposite directions — and those cases are common enough that they are the reason for this article.

The identity that explains everything: SWOLF = time × (1 + stroke rate)

Let t be the length time in seconds and n the stroke count. Your stroke rate f, in strokes per second, is n ÷ t, so n = f·t. Substitute into SWOLF = t + n and you get SWOLF = t + f·t = t·(1 + f). Check it: 30 seconds at 20 strokes is a rate of 0.6667 strokes per second, and 30 × 1.6667 = 50. Thirty-two seconds at 17 strokes is 0.53125 per second, and 32 × 1.53125 = 49 exactly.

That factorisation is worth more than the definition, because it separates the two things SWOLF confuses. The first factor is pure speed: hold the stroke rate constant and SWOLF moves exactly in proportion to your time. The second factor is pure technique: hold the time constant and SWOLF moves exactly in proportion to your stroke rate. A score that changed can always be decomposed into how much of it was the clock and how much was the arms, and the two levers are entirely independent of each other.

Trap one: it rewards gliding, at one stroke per second

Because SWOLF is a plain sum, its exchange rate is exactly one stroke for one second. Any change that removes a stroke while adding less than a second improves the score. That is a real technical fact and often good advice — lengthening the glide, tightening the catch and fixing a crossover all remove strokes without costing much time. But it is also a loophole, and it is easy to fall into without meaning to.

Work the three 25 m swims in the table. The baseline is 30.0 seconds in 20 strokes, a score of 50. Extend the glide and it becomes 32.0 seconds in 17 strokes: three strokes saved for two seconds added, so the score falls to 49. That is an improvement by the only rule SWOLF knows. But at 32 seconds a length, a 1,500 m swim takes 60 x 32 = 1,920 seconds, or 32:00, against 60 x 30 = 1,800 seconds, or 30:00. The score got better and the swim got two minutes slower.

The mirror case is just as instructive. Raise the stroke rate instead — 28.0 seconds in 23 strokes — and the score rises to 51, the worst of the three, while the 1,500 m falls to 28:00, the fastest of the three. Rank the three swims by SWOLF and you get exactly the reverse of ranking them by time. A metric that inverts the ordering of the thing you care about is not broken, but it is certainly not a proxy for it, and anyone chasing a lower number for its own sake is training themselves to swim slowly with beautiful arms.

Trap two: SWOLF is proportional to the pool

This one can be proved rather than observed. For a length of L metres swum at velocity v with a distance per stroke of d, the time is t = L ÷ v and the stroke count is n = L ÷ d. So SWOLF = L ÷ v + L ÷ d = L × (1 ÷ v + 1 ÷ d). The pool length factors straight out of the front. Double the pool and, for identical swimming, you double the score — it is not an approximation, it is an exact proportionality.

Our baseline swimmer covers 25 m at 0.833 m/s with 1.25 m per stroke, giving a score of 50. Feed L = 50 into the same formula with the same v and d and it returns exactly 100. Real life adds a little on top, because a 50 m pool gives you half as many push-offs per hundred metres: the table's fourth row has her at 62.0 seconds and 42 strokes, a score of 104, and a 1,500 m time of 31:00 against 30:00 in the short course. Same swimmer, same technique, same week. A SWOLF of 50 and a SWOLF of 104 are the same performance.

There is a subtler version of the same problem that catches swimmers in North America. A 25 yard pool is 22.86 m, which is 8.56% shorter than 25 m. Scale the baseline swim by that factor and the length takes 27.4 seconds and 18 strokes, for a SWOLF of about 46. The same swimmer, the same stroke, on the same day, reads 50 in one pool and 46 in the other — a four-point improvement bought entirely by changing buildings. If your training log spans both, it is recording the pool and not you.

The alternatives, and why they do not fully fix it

The dimensionally honest cousin is the stroke index, defined by Costill and colleagues as velocity multiplied by distance per stroke. It has real units of square metres per second, higher is better, and it is not tied to pool length. It is a better-behaved quantity in every formal sense. But run our three swims through it and the result is uncomfortable: the baseline scores 1.042, the gliding swim 1.149 and the fast swim 0.971. The gliding swim wins again. Fixing the dimensions did not fix the incentive, because both metrics reward distance per stroke and neither of them knows you were in a race.

The real fix is not a better single number, it is refusing to collapse two variables into one. Log the time and the stroke count separately and both traps disappear. You can then see the actual question — am I holding this pace with fewer strokes than last month? — instead of inferring it from a sum that has already answered it for you with an exchange rate nobody chose.

How to actually use it

Three conditions turn SWOLF from a decoration into an instrument, and all three are about holding things still. Fix the pool: one length, one venue, or the number is measuring architecture. Fix the pace: swim your test lengths on a set send-off, so that the time factor in t·(1+f) is constant and every movement in the score is the stroke rate. Fix the stroke: a SWOLF from front crawl and one from breaststroke have nothing to say to each other.

Then read changes, never levels. There is no good absolute SWOLF, because the number embeds your pool length, your height, your arm span and the stroke you chose; comparing yours to a stranger's on a forum is meaningless in the strictest sense. Comparing this Tuesday's to last Tuesday's, at the same send-off in the same pool, is one of the cleanest technique signals an amateur swimmer can get for free — because under those conditions the only thing that can have moved is the number of strokes it took you.

Frequently asked questions

What is a good SWOLF score?

The question has no answer that survives contact with the arithmetic. The score is proportional to pool length, sensitive to your height and arm span, different for every stroke and dependent on the pace you happened to swim. Published bands exist and are mostly repackaged from short-course front crawl; treating them as targets encourages exactly the gliding behaviour that makes the metric lie. Your own baseline, measured under fixed conditions, is the only benchmark worth having.

My watch shows my SWOLF improving but my times are not. What is happening?

You are almost certainly gliding more. Check the two components rather than the sum: if strokes per length are falling while seconds per length are rising, the score is improving for the wrong reason. Whether that is a problem depends on the season. In a technique block it is exactly what you asked for. Four weeks before a race it is the opposite of what you want, and the fix is to hold a fixed pace and try to take strokes out at that pace instead of at any pace.

Should I count the push-off as part of the length?

Your watch does, because it times wall to wall and counts every arm cycle in between, and you should do the same by hand for consistency. It is worth knowing what that includes, though: a long underwater phase reduces both the time and the stroke count for the same length, so it improves SWOLF twice over. That is not cheating — a good push-off is genuinely fast — but it means a change in your turns can move the score without anything happening to your actual swimming.

Does SWOLF work for open water?

Not as defined, because there are no lengths. Watches substitute a fixed distance, usually 25 m, and estimate the strokes over it, which reintroduces every problem at once: the distance comes from GPS smoothed through a moving arm, and current, chop and sighting all move the time without touching your technique. In open water, distance per stroke and stroke rate logged separately are far more informative, and honest about their own uncertainty in a way a single fused score is not.

How many lengths should a SWOLF test be?

Enough that a single miscount does not decide the result, and few enough that fatigue does not. A practical protocol is six to ten lengths on a fixed send-off after a full warm-up, discarding the first as a settling length and averaging the rest. Repeat it in the same session every two to three weeks, at the same time of day and at the same pace. Under those conditions a change of two points is worth reading, and a change of one is noise.