While compiling data for the 12-meter vs 20-meter draft-zone discussion, I went down a bit of a data rabbit hole lol so I thought I would post it in here, especially since there are a lot of data-minded folks here.
First, just to be clear up front: I’m not posting this to argue that 20m is right or that 12m is right. I genuinely don’t have a conclusion here. This is more about sharing some numbers I pulled together and seeing if others notice anything I might be missing.
Initially, I was able to get access to Ironman World Championship data going back to 2012, looking at out-of-the-water pack sizes and pack sizes at Hawi (turnaround). That data ended up being pretty inconclusive—2016 in particular had a massive pack that kind of skewed things so I didn’t feel confident drawing anything meaningful from that alone. (if you want me to post it i will)
From there, I shifted gears and compiled data from:
I fully recognize this is very much apples to oranges:
-
Different field sizes
-
Different course profiles
-
Different athlete dynamics
-
Different race incentives
So direct comparisons are inherently flawed. That said, I wasn’t trying to “prove” anything more just looking to see if any patterns emerged at all.
What I tracked was fairly simple:
Give or take a bit on timing and distance, but I tried to keep it reasonably consistent.
Honestly, I didn’t walk away with a strong takeaway. I even ran the dataset through ChatGPT Pro to see if it could identify anything meaningful. It spit out a few possible trends, but nothing definitive—some of it mildly interesting, nothing conclusive. I’ll post both the table and the pattern observations below so people can judge for themselves.
The main reason I’m sharing this here is curiosity:
If anything, this exercise mostly reinforced the idea that controlled theory-testing only goes so far, and that real-world testing is probably the only way to get meaningful answers.
I really liked what @E_DUB mentioned in his recent article: that a logical next step might be for IRONMAN to test a 20-meter draft zone at smaller 70.3 Pro races, where variables are easier to manage, and outcomes might be clearer.
Anyway, here’s the data.
Do with it what you want.
Genuinely curious to hear your thoughts.
Chatgpt respose:
Here’s what jumps out to me when you treat this like a “pack physics” dataset (not just anecdotes), while keeping in mind T100 = 20 starters, IM = ~70–75 starters.
-
The biggest signal: 20m races shrink packs; 12m races often grow packs
If you compare largest swim-exit group → largest group around ~35k, you get two very different behaviors:
T100 (20m drafting)
• Swim-exit “largest group” is often huge in percentage terms (because the fields are small and swim parity is high): examples: French Riviera 15/20, Spain 14/20, Wollongong 14/20, Qatar 11/20
• But by ~35k, the largest group is usually only 2–6 riders:
◦ Singapore: 9 → 3
◦ Vancouver: 8 → 4
◦ French Riviera: 15 → 4
◦ Spain: 14 → 4
◦ London: 7 → 2 (max, with multiple 2-man packs)
Key pattern: in your T100 rows, the biggest group never gets bigger on the bike. It always fractures.
Ironman (12m drafting)
In multiple IM races, the biggest group at ~35k is as big or bigger than the biggest swim-exit group:
• IM Texas: 8 → 15
• Frankfurt: 4 → 9
• Nice WC: 10 → 14
• (Even when the swim group is modest, the bike group can still become large.)
Key pattern: in your IM rows, the bike often coalesces athletes into a larger legal pack.
That “coalescence vs fragmentation” difference is the cleanest signal in what you collected.
-
Absolute pack size matters more than percent-of-field for “race feel”
Because T100 is capped at 20, a 6-person pack is already 30% of the field—but it’s still six riders. In Ironman, a 9–15 person group is a completely different animal in terms of:
• how easy it is to “sit in”
• how sticky the pack becomes (more wheels, more shelter, more elastic)
• how hard it is to officiate in real time
• how hard it is for a single rider to break it
So even if some of the percentages don’t look wildly different on paper, the absolute group size does.
-
Your dataset suggests a “pack cap” effect at 20m
T100 mid-bike largest groups cluster around 4 (with a range of 2–6). That looks like a practical cap that emerges when:
• the distance is long enough that sustained surging matters,
• but the draft benefit is reduced and the “accordion” costs more,
• and moving up / staying legal requires more power and/or space.
Whereas at 12m, you’re repeatedly seeing 9–15 as a totally normal mid-bike “largest group” outcome.
-
Swim parity vs drafting rule: you can separate them with what you already captured
T100 swim groups are big mainly because:
• smaller field
• very high swim parity in T100
• (often) wetsuit/non-wetsuit dynamics and course specifics
But the rule shows up in what happens after that:
• In T100, huge swim packs do not survive to 35k as big packs.
• In IM, smaller swim packs can merge into bigger bike packs.
That’s exactly what you’d expect if 12m encourages “pack formation” and 20m discourages it.
-
The one “warning flag”: course + conditions can mimic either outcome
Spain 70.3 WC is a good example: you had a massive early separation (26 in 40s in a “middle” group), but by ~35k the largest is ~10. That shows you can get fragmentation at 12m if:
• the course forces selection (climbs/technical)
• wind creates splits
• athletes light it up early
• officiating is tight
So your data supports: draft distance is a strong lever, but not the only lever.
What I’d say as the honest “data-driven takeaway”
From this table alone, the most defensible statement is:
20m (T100) appears to reduce pack persistence and pack coalescence. Big swim groups break into smaller bike groups by ~35k, and the “largest group” tends to stabilize around ~2–6 riders. 12m (Ironman) appears to allow pack coalescence. Even when swim groups are smaller, groups can grow into 9–15+ on the bike by ~35k.
That’s not “proof” in a scientific sense (small sample, mixed courses, mixed formats), but it’s a consistent directional signal across your rows.
