Power v Giants: opposition report and game plan

Performance analysis, prepared for the senior coach. Source: the season tracking database (244 games: the Giants' and the Power's 112 games each against 28 common opponents, plus 20 head-to-heads), the league position rating tool, and the league match engine. Every number below comes from that data. The working notebook (notebook.md) and scripts (work/) hold the detail.

The short version

  1. The Giants kick, mark and kick again, and almost never handball (0.8 handballs a game; 98% of their marks are kicked straight away, 0.4 s after the mark). Up front they hold two forwards out very wide and one central target, and feed the wide men into the middle. High confidence
  2. Without the ball they sit off and intercept. They put less pressure on the ball carrier than anyone (0.55 in their back 50, against 1.35 from the league against them) and read the kick: their three defenders intercept-marked 13.8% of the Power's kicks in our 20 meetings (8.5% for other clubs). Intercept marks gave them 27% of their points against us. High confidence
  3. Their kicking breaks down under pressure. In their back third against us they kept 80% of kicks unpressured and 62% under pressure. Because they never handball, they have no outlet. Our wins came with more pressure (correlation with our margin r = +0.52) and fewer Giants intercepts (2.6 a game in wins, 3.9 in losses). Medium-high confidence
  4. They run themselves into the ground. Their players spend 55-59% of the time above sprint pace (Power 34-36%). By the second minute their midfielders are at about a quarter of full energy and their forwards at about a third. We win 58% of contested ground balls against them. Medium-high confidence
  5. We play into their hands in two ways. (a) Our defenders punch the ball away even when no Giant is there: 3.2 uncontested spoils a game. (b) We kick long into the corridor from set plays: 23% of those are intercepted. Both are easy to fix. Medium-high confidence

Game plan in one line: press their kicker with our (fresh) forwards, mark the ball when alone, stop feeding their interceptors with set-play kicks into the corridor and short chips, and keep the ball on the ground where their tired midfield loses. We are already modest favourites: 11 wins, 1 draw, 8 losses, +4.0 points a game. The aim is to remove the games we lose.

1. Recommendations: how the Power should play the Giants

  1. Press the Giants' kicker, especially in their back half, and use the forwards to do it. Medium-high confidence
    What to change: today only our midfielders tackle: all 212 of our tackles this season (182 against others, 30 against the Giants) were by m1/m2, and our forwards sprint only 16% of the time. Give the nearest forward (or midfielder) the job of closing the Giants' ball carrier during their 0.4 s gather. This matters most on their back-third kicks, about 6 a game of which are currently unpressured.
    Why the data says it works: Giants kick retention falls from 70% unpressured to 55% under heavy pressure (44% against us). Our own retention does not fall, because we handball. The league rating tool, applied to 163 real head-to-head frames, values putting our nearest player on their carrier at +1.4 percentage points of "Power scores the next goal" each time (positive in 90% of cases). That is after charging for the man left free. In the match engine, one extra presser on 91 real Giants kicks cut their retention from 52.6% to 48.5%, and our win-backs rose from 30% to 35%. Our forwards have the energy budget: estimated 0.80 against the Giants' defenders' 0.60. And a chain that starts in our forward 35 is worth 1.7-2.2 points.
    Risk: a pressing forward leaves a Giants defender free. The rating test already includes that cost, and the Giants' back-half chains are their least productive (0.12 points per chain).
  2. Mark it when you are alone. Spoil only when a Giant is there. Medium-high confidence
    What to change: half of our spoils against the Giants are uncontested (3.2 a game). A spoil gives back a loose ball we regain 85% of the time; a mark is a protected set play we keep about 95% of the time.
    Why: on the 65 real uncontested head-to-head spoils, the rating tool scores "our defender holding the ball" +5.1 pp above the loose ball that actually followed (95% of cases positive). Defenders are strong marks by the league's own player profile. When a Giant is there, keep spoiling to ground: we win 58% of contested ground balls against them.
  3. Stop feeding their interceptors: no long set-play kicks into the corridor, and no short chips. Medium confidence
    What to change: from a mark, either play on quickly (handball or run), or kick only to a forward who has broken clear (no Giant within ~10 m), usually 22-35 m away. Drop the 15-22 m chip (about 30% of our general-play kicks).
    Why: the Giants intercept 23% of our set-play kicks and 21% of our unpressured kicks (6% when we kick under pressure). Their defenders set up goal-side while our kicker takes his time. In the engine, our kicks to a teammate with no Giant within 10 m were kept 73% of the time, against 54-59% when a Giant was within 10 m. A free forward 22-40 m away was available in 53% of situations. Across the whole season, our sub-22 m kicks start chains worth 0.52 points (from the defensive midfield), against 1.03-1.49 for longer kicks.
    How: our forwards are almost static (16% sprint). Timed leads into space are what create the free target. That is my inference, not directly measured.
  4. Make their midfield run, and keep the ball low. Medium confidence
    Why: the Giants' midfielders sit at an estimated 0.25 energy from the second minute (about 62% of top pace); ours at 0.37-0.39. When two players reach a loose ball at similar distance with an energy gap above 0.2, the fresher one wins 58% (1,686 cases). The Giants score least in the third minute against us (1.75 points a game). We score 4.0 points a game in each of minutes 2 and 3 against them (4.8 in minute 3 against others). Avoid floating kicks to contests where their defenders mark. Ground-ball contests and quick handball chains suit us.
  5. Defend the Giants' forward pattern: f1/f3 wide into f2 central. Medium confidence
    Why: 43% of their f1's kicks and 38% of their f3's go to the central f2, and f2 kicks out to the wide men 60% of the time. Our d2 is the key player: stay goal-side of their f2 and let the wide feed come. Against us, Giants kicks from their forward midfield (70-105 m) are already kept only 49% of the time, and their goal conversion is poor (29% in general, 31% against us). Keep contesting the shooter: 36% of their shots come under heavy pressure.
  6. Centre bounces: keep doing what we do. High confidence
    The m1 v m1 bounce is a coin flip with identical players (we win 55% against them). Energy makes no difference (r = 0.02). Our clearance after winning it is the best part of our game against them: kept 78% of the time, and we score next 65% of the time. Do not change it.

2. How the Giants play

Figure 1. Average shape, Giants (orange) v Power (blue), in games against common opponents. In each frame every line is sorted left / centre / right, so width is visible. The Giants' two wide forwards hold near the boundary; the Power are narrow.

With the ball

GiantsPowerLeague (v Giants)
Handballs per game0.88.65.9
Marks per game8.75.69.6
Set plays kicked (v played on)98%, 0.4 s after the mark58%82%
Average kick (non-shot)29.3 m, gain 18.7 m26.3 m, gain 21.1 m28.7 m, gain 19.1 m
Mean landing width |y|16.4 m11.0 m15.6 m
Kicks retained62%64%66%
Shots, goal %5.8 a game, 29%5.8, 39%7.1, 33%
Figure 2. Where kicks land (each club attacking left to right, v common opponents). The Giants use the whole width; the Power kick down the middle.
Figure 3. Who receives each player's kicks (% of that kicker's kicks, v common opponents). Note the Giants' f1→f2 (43%), f3→f2 (38%) and f2→f1/f3 (60%) triangle.

Without the ball

Figure 4. Left: share of kicks retained by pressure on the kicker. The Giants' retention falls away; the Power's does not, because they handball out of trouble. Right: pressure applied to the opposition carrier by zone. The Giants sit off in their back half.

Stoppages and transition

Figure 5. Who scores next after a centre bounce.

3. Where the Giants are vulnerable

WeaknessEvidenceConfidence
Kicking under pressure, with no handball outletRetention 70% → 55% (v us: 68-72% → 44%); back-third kicks v us 80% → 62%. Our margin rises with the pressure we apply (r = 0.52, n = 20).Medium-high confidence
Fatigue: midfield and forwardsSprint share 55-59%; estimated energy mids 0.25, forwards 0.36 from minute 2; top-end speed 0.71-0.73 of maximum after minute 1 (Power 0.80-0.82); we win 58% of contested ground balls.Medium-high confidence
Poor finishing29% goals from shots (v others) and 23% from 20-30 m (Power 42%); 36% of their shots under heavy pressure, 36% from angles wider than 35°.High confidence
Little scoring from their back halfChains starting in their back 35 m are worth 0.12 points (Power 0.34). Locking the ball in our forward half costs them a lot.Medium confidence
Long, direct opponentsClubs whose kicks gain more ground do relatively better against the Giants than against us (r = 0.53 across 27 clubs). The Demons, Cats and Warriors are the clearest cases. This is a correlation across clubs with only 4 games each.Low-medium confidence
Figure 6. Left: estimated energy by line over the game (from tracking speeds and the league's published energy law; energy itself is not in the data). Right: points scored in each minute.
Figure 7. Shot locations and outcomes against common opponents.
Figure 8. The 27 common opponents (the Blues have no non-shot kicks recorded and are left out): forward gain per kick against how much better they do against the Giants than against the Power.

4. How the Power play now, and why the head-to-heads go each way

Our style has been stable all season (handballs 8.6 a game in both halves of the season; tackles 1.9 and 1.3). In short:

Figure 9. The 20 head-to-heads. Right: our margin against the pressure we put on Giants carriers (bubble size = Giants intercept marks that game).
Average per gamePower wins (11)Giants wins (8)
Power points per chain1.050.55
Giants intercept marks2.63.9
Giants kick retention57%63%
Power pressure on Giants carrier1.181.07
Power goal accuracy48%24%
Centre bounces won by Power52%56%

Why the Giants beat us: they intercept our kicks and turn them into scores, and on those days we also kick badly at goal. Why we beat them: we press their kicker, they turn it over, and we win the ground ball. Twenty games is a small sample: the correlations (pressure +0.52, Giants intercepts −0.42, Giants retention −0.52) point the same way as the larger season-wide evidence, but on their own they would be weak.

Figure 10. Our kicks against the Giants (15 m+). Left: landing points, with the Giants' intercept marks in red. Right: intercept rate by the kicker's pressure, set play or not, and landing lane.
Figure 11. A typical case (game 227, 2:04): a set-play kick into the corridor, read by the Giants' d3, who is already goal-side of our target.

5. Testing the ideas

Three kinds of test were run. All of them use real head-to-head positions.

Figure 12. Left: rating-tool counterfactuals (error bars = 95% interval). Centre: engine test of one extra presser on the Giants' kicker. Right: engine test of our kick options by length and by how much space the target has.
IdeaResultVerdict
Press their kickerRating +1.4 pp per instance; engine −4.1 pp Giants retention (SE 2.2); match data −15 pp at heavy pressureRecommended
Mark, don't spoil, when aloneRating +5.1 pp per instance (n = 65); regain 85% (spoil) v ~95% (mark)Recommended
Kick to a free forward 22-35 m away; drop the short chipEngine: 73% (free) v 54-59% (Giant within 10 m); short kicks start chains worth half as muchRecommended
Kick wide rather than down the corridorIntercepts lower wide (3% v 16%) but small sample (65 wide kicks); the rating tool slightly prefers central; value change flatNot recommended: inconclusive
Tilt the centre bounce through m1 energyNo relationship (r = 0.02, p = 0.42)Rejected
Attack a high Giants defensive lineTheir line is not high (x 68-70 v the league's 78-82 with the ball in their forward third)Rejected

6. What I could not establish, and how far to trust this

Method notes. All positions are converted to each club's frame (attacking left to right, a 180° rotation for away games). Possession chains are runs of control by one club. Pressure uses the league formula (sum of exp(−(d/5)²), capped at 2). "League v Giants" means the 28 common opponents in their games against the Giants. Scripts: work/*.py; figures: fig/; notebook: notebook.md.