AMP Analytics Arena

Model by KSplit

Model Diagnostics — Live Season Log
Games Logged 4010
Updates in real time
Total Games Logged
4010
season to date
MAE (Ks)
1.756
median absolute error
Avg Signed Error
0.040
direction of bias
Within ±1K Rate
50.32%
within ±2K: 74.54%
01 — Ceiling Profile
How Do Ceiling Profile Labels Perform?
Tracks how often each ceiling profile delivers upside signal. High | Tail-Driven games show the highest 2+ over rates as the log grows — this table confirms or challenges that signal in real time.
Low | Centered
n = 2015 games
2+ Ks Over Line
15.88%
Actual Ks Over the Line
44.96%
2+ Ks Under Line
36.18%
Mid | Tail-Supported
n = 1525 games
2+ Ks Over Line
21.18%
Actual Ks Over the Line
51.28%
2+ Ks Under Line
32.13%
High | Tail-Driven
n = 470 games
2+ Ks Over Line
21.28%
Actual Ks Over the Line
55.11%
2+ Ks Under Line
31.70%
02 — Right Tail Calibration
Mean-Based Right Tail Tables
When the model assigns a given upside probability, how often does it actually happen? O/E near 1.00 means the model is well-calibrated at that bucket. Values above 1.0 mean reality exceeds the model's prediction — below 1.0 means the model is over-calling it.
+1 tail calibration (bucketed by P(+1))
P(+1) bucketHitsnHit %
<30%420145928.79%
30–39.99%496152832.46%
40–49.99%30084735.42%
50–59.99%5615037.33%
60–69.99%62326.09%
70%+020.00%
Total Games4009
03 — Tail Environment
Strikeout Environment Classification
Environment type tracks the structural conditions around each game's K distribution.
Environment TypeSample SizeActual Ks: 2+ Ks Over the LineActual Ks Exactly 1 Over LineActual Ks Exactly 1 Under LineActual Ks: 2+ Ks Under the Line
Fragile Tail Environment193115.59%13.10%19.21%35.84%
Low Tail Environment8023.75%7.50%21.25%31.25%
Balanced Tail Environment102620.57%13.55%17.74%30.41%
Balanced Tail (Accessible)50321.87%13.72%17.50%32.60%
Tail-Friendly Environment28820.14%13.89%13.19%34.38%
Stable Tail Environment18224.18%15.38%15.93%25.27%
All Environments4010
04 — Median by Tier
Exact Hits by Pitcher Tier
Exact hits become less frequent as distributions widen in higher strikeout tiers.
Elite
n = 499 games
MAE
1.962
Median Exact Hit
17.84%
Within ±1
44.47%
Within ±2
68.41%
Mid
n = 1161 games
MAE
1.849
Median Exact Hit
17.48%
Within ±1
48.65%
Within ±2
70.20%
Low
n = 2379 games
MAE
1.667
Median Exact Hit
17.91%
Within ±1
52.35%
Within ±2
77.93%
05 — Median Diagnostics & Edge Band
Error Distribution & ROI by Edge Bucket
Error BucketsMedian
-2K (From Actual)12.47%
-1K17.53%
Exact Hit17.91%
+1K14.89%
+2K11.75%
Big Miss % ABS Err>2.5)25.46%
Edge Band (Best Edge)Sample SizeROI (Units / Bet)
< 0%00.00
0–4.99%16540.00
5–9.99%12900.00
10–14.99%6880.03
15–19.99%263-0.04
20–24.99%81-0.11
25–29.99%30-0.02
30%+4-0.09
06 — Over/Under Edge Band Performance
ROI & Unit P/L by Edge Band Bucket
Overs
Edge Band (Overs)Sample SizeROI (Units / Bet)Unit P/L
0–4.99%874-0.01-8.84
5–9.99%713-0.04-29.45
10–14.99%3540.0311.68
15–19.99%122-0.04-5.06
20–24.99%40-0.15-6.03
25–29.99%12-0.20-2.39
30%+3-0.46-1.38
Total2118-0.02-41.47
Unders
Edge Band (Unders)Sample SizeROI (Units / Bet)Unit P/L
0–4.99%7800.001.11
5–9.99%5770.0631.81
10–14.99%3340.0310.16
15–19.99%141-0.05-6.77
20–24.99%41-0.06-2.57
25–29.99%180.091.66
30%+11.021.02
Total18920.0236.42
07 — Rolling Window Summary
Combined P/L Across Edge Bands
L200 All
Edge BandUnit P/L
0–4.99%-7.75
5–9.99%-4.10
10–14.99%0.97
15–19.99%-1.40
20–24.99%0.00
25–29.99%0.00
30%+0
L200 Total-12.28
L50 All
Edge BandUnit P/L
0–4.99%-7.51
5–9.99%-5.53
10–14.99%0.15
15–19.99%1.94
20–24.99%0.00
25–29.99%0
30%+0
L50 Total-10.95
KSplitAnalytics.com
Live season log · Updates as games are recorded

Today's Board · Pitcher Distributions