REGULAR SEASON2026-10-17
WSH VS NJD13:00 ET
THE LEONARD EFFECT
CAPITALS ANALYTICS
COMMAND CENTER
Original TLE tools for Washington hockey. Fan mode for the picture. Analyst mode for the wiring.
Updated Oct 16, 2026, 8:22 AM America/New_York · Advanced analytics pending for the latest game.
No published Capitals analytics yet.
No published Capitals analytics yet. The desk will light up when a game is published.
Game DNA
Select a published game. Unpublished schedule nights stay pending and are not treated as results.
Pressure Map
Weights live in newsroom settings and are disclosed on Methodology. Washington events push the wave up; opponent events push it down.
What Changed
Deltas versus the prior published sample, with sample size attached. Nothing here is invented.
Rolling picture
Click a game to open Game Lab. Rolling 3 / 5 / 10 / 20 / full use published games ordered by date.
Locker room
Names and sweater numbers as text. No headshots. Five-season counting stats when loaded; on-ice sample is published games.
Ryan Leonard
LW · R
Pierre-Luc Dubois
C · L
Alex Tuch
RW · R
Alex Ovechkin
LW · R
Dylan Strome
C · L
Jordan Kyrou
RW · R
Aliaksei Protas
LW · L
Ilya Protas
C · L
Tom Wilson
RW · R
Boone Jenner
LW · L
Justin Sourdif
C · R
Anthony Beauvillier
RW · L
Jakob Chychrun
LD · L
Matt Roy
RD · R
Martin Fehérváry
LD · L
Timothy Liljegren
RD · R
Cole Hutson
LD · L
Vincent Desharnais
RD · R
Logan Thompson
G · R
Charlie Lindgren
G · R
Ivan Miroshnichenko
LW · R
AHL
Ethen Frank
RW · R
AHL
Bogdan Trineyev
RW · R
AHL
Andrew Cristall
LW · L
AHL
Clay Stevenson
G · L
AHL
Jonny Brodzinski
C · R
Josh Dunne
C · L
Rasmus Sandin
LD · L
Dylan McIlrath
RD · R
Combinations, chemistry, matchups
Projected 2026-27 paper lines (RMNB / public reporting). Camp will move these. Not official.
Desktop: drag a skater onto a slot. Phone: tap a skater, then tap a slot. Matching position is preferred; off-wing is allowed. No duplicates inside a unit group.
F1
F2
F3
F4
D1
D2
D3
PP1
PP2
PK1
PK2
Chemistry Matrix
Matchup Fingerprint
Unavailable until opponent analytics are entered. TLE will not invent the other bench.
Original TLE rink
No shot locations entered for this filter.
Saves versus expected
PP / PK
Efficiency and what-if
Opportunity, drops, regression
Drop Detector
Regression Watch
Direction of travel
How the Command Center thinks
Formula versions: fantasy v1. Pressure weights attempts 0.25, xG 0.35, HD 0.25, PP 0.15. Last updated Fri Aug 28 2026 20:56:06 GMT+0000 (Coordinated Universal Time).
- Corsi For (CF)
CF = SOG + missed shots + blocked shots when those columns exist; otherwise the stored CF value.
Attempt counts without shot-quality context. - Fenwick For (FF)
FF = SOG + missed shots (or stored FF).
Still unweighted for danger. - CF%
CF / (CF + CA) × 100
Share stats need sample confidence. - FF%
FF / (FF + FA) × 100
- GF%
GF / (GF + GA) × 100
Results-heavy and noisy in small samples. - xGF%
xGF / (xGF + xGA) × 100, only when both sides are present.
xG is Unavailable unless imported or entered with source and model_version. TLE does not claim a proprietary xG model. - PDO
On-ice SH% + on-ice SV% on a 1.000 scale (shots for/against).
PDO is a luck index. Warnings fire above 1.05 on n<5. Never treated as certainty. - GSAx
xGA − GA. Positive means above expected saves.
Requires entered xGA. - Per 60
value / (toi_seconds / 3600)
Unstable below Medium sample confidence. - Rolling windows
L5 / L10 / L20 from published games ordered by date. Season totals are always SUM of published games, never a separately stored season row.
- Sample confidence
High ≥ 300 minutes, Medium ≥ 80, Low ≥ 20, else Very Low.
Minutes are TOI, not wall-clock. - TLE Pressure Map
Configurable weights stored in analytics_settings JSON. Defaults: attempts 0.25, xG 0.35, HD 0.25, PP 0.15. Washington events are positive; opponent events are negative.
A TLE visualization, not a proprietary expected-goals engine. - Line projection
Shared TOI ≥ 300 min: actual line CF%/xGF% (High). ≥ 80 min: 70% together + 30% harmonic mean of individuals (Medium). Else individuals only (Low). Explanation always includes minutes together.
Does not invent opponent matchup data. - Fantasy Opportunity Score v1
clip(0.25·toi_delta + 0.2·shot_rate_delta + 0.2·ixg_l5 + 0.15·pp1_flag·20 + 0.1·upcoming_density + 0.1·hd_rate, 0, 100)
Will not dump an elite career pts/60 skater for three bad games. Versioned; current version is v1. - Regression Watch
Compares G vs ixG, SH% vs baseline, and PDO. Labels: Possible Positive / Approximately Sustainable / Possible Negative.
Labels are never certainty. - Cap math
Hit = sum of published (non-dead) contracts for the season. Space = upper limit − hit. Upper/floor come from analytics_cap_limits. What-if clones the roster in memory and is labeled SCENARIO.
Unofficial tracking, not a league salary-cap system.
Limitations
- Public pages read published games only. Drafts stay in the newsroom.
- No MoneyPuck, Natural Stat Trick, or other proprietary feeds are scraped.
- League ranks stay an em dash unless league-wide data is entered.
- Microstats hide in the public UI when every microstat column is null.
- Independent unofficial fan publication. No NHL or Capitals marks as decoration.
The Leonard Effect is an independent unofficial fan publication. Not affiliated with, endorsed by, or sponsored by the NHL, the Washington Capitals, or the NHLPA. Player names appear as factual text. No official logos, headshots, or trademarks are used as decoration.