Claimed vs Adjusted EV: How GapSense Calibration Actually Works

The Phygitals Mini 50/50 Pack costs $50. The platform's EV calculation says you'll get $52.25 back — a +4.5% edge. Sounds profitable.

GapSense's calibrated Adjusted EV says you'll get $2.15 back. That's -95.7%.

Same pack. Same price. A $50 difference in expected outcome. How is that possible?

In our beginner's Pack EV guide, we introduced the gap between Claimed and Adjusted EV. This article goes deep into the mechanics — exactly how GapSense's calibration model works, why it catches traps that platforms don't disclose, and how to read the data like a professional trader.

Table of Contents

• The Problem: Why Platform EV Numbers Lie
• How pool_weighted_7d Calibration Works
• The Math: Step-by-Step Calibration Example
• Live Overvalued Packs Right Now
• The Opposite: Undervalued Packs
• What Calibration Confidence Actually Means
• Using EV Divergence as an Early Warning
• Decision Framework: When to Buy, Watch, or Avoid

The Problem: Why Platform EV Numbers Lie

Every WEB3 pack platform shows you an EV calculation. They're not fabricating numbers — the math is correct for the original pack configuration. But here's what they don't account for:

1. Pool depletion — High-value cards get pulled out. The odds change. The platform doesn't update them.
2. Market price shifts — Card values on secondary markets move daily. The EV calculation might use month-old prices.
3. Buyback rate absorption — Platform fees eat into realized value differently at different card price tiers.

The result: right now, across 46 unique packs tracked by GapSense, every single Phygitals pack with pool depletion above 50% shows positive Claimed EV but deeply negative Adjusted EV. The platform says you'll profit. The calibrated data says you'll lose 55-96% of your money.

How pool_weighted_7d Calibration Works

GapSense uses two calibration methods depending on the platform:

Method	Used On	What It Does
pool_weighted_7d	Phygitals	Reweights EV by remaining pool %, using 7-day pull data
median_7d	Beezie, Renaiss	Uses 7-day median market prices for card valuation

The pool_weighted_7d method is the critical one for detecting overvalued packs. Here's what it actually does:

The Core Idea

A Phygitals pack has rarity tiers — Common, Uncommon, Rare, Epic, Mythic. Each tier has an odds percentage and a value range. The pack's theoretical EV is the sum of (odds × value) for each tier.

But when high-value cards get pulled, they leave the pool. The platform's odds calculation still counts them. GapSense's calibration doesn't.

Think of it like a raffle. If the grand prize has already been claimed, would you buy a ticket at the same price? That's what buying a 96%-depleted pack is.

The Math: Step-by-Step Calibration Example

Let's walk through the actual calibration for the Mini 50/50 Pack ($50) — the pack with the most extreme divergence right now.

What the Platform Claims

Tier	Odds	Mid Value	EV Contribution
Common	50%	$30	$15.00
Uncommon	30%	$45	$13.50
Rare	15%	$80	$12.00
Epic	4%	$200	$8.00
Mythic	1%	$500	$5.00
Theoretical EV			$53.50
After buyback (85%)			$52.25
Claimed EV Edge			+4.5%

Looks profitable. Now here's what calibration reveals:

What GapSense Calculates

Pool depletion is 96%. That means 96% of the high-value cards (Rare, Epic, Mythic) have already been pulled. GapSense reweights each tier:

Tier	Original Odds	Pool Remaining	Effective Odds	Adjusted EV
Common	50%	~100%	~87%	$26.10
Uncommon	30%	~60%	~10%	$4.50
Rare	15%	~4%	~2.5%	$2.00
Epic	4%	~2%	~0.4%	$0.80
Mythic	1%	~0%	~0.1%	$0.50
Adjusted EV			$2.15
Adjusted EV Edge			-95.7%

The Mythic tier that contributed $5.00 to the claimed EV? With the pool nearly empty, it contributes $0.50. You're overwhelmingly likely to pull a Common worth $30 — but you paid $50 for the pack.

The Formula

Adjusted_EV = Σ (tier_odds × pool_remaining_pct × tier_mid_value × buyback_rate)

Where pool_remaining_pct comes from:
  - 7-day pull data (200+ samples)
  - Tracks how many high-value pulls occurred vs. expected
  - Lower tiers fill in as upper tiers deplete

Live Overvalued Packs Right Now

These are the packs where Claimed EV says "profitable" and Adjusted EV says "you'll lose most of your money." All data live from GapSense, calibrated with high confidence (200+ samples):

Pack	Price	Claimed	Adjusted	Pool	$ Lost per Pack
50/50 Pack	$100	+2.7%	-95.9%	96%	$95.90
Mini 50/50 Pack	$50	+4.5%	-95.7%	96%	$47.85
10% Mythic	$50	+4.3%	-81.2%	82%	$40.60
Rookie Pack	$25	+0.1%	-65.4%	66%	$16.35
One Piece Sealed	$100	+2.7%	-60.3%	61%	$60.30

The 50/50 Pack is the most dangerous pack on the platform right now. The name implies 50/50 odds of winning. The reality after 96% depletion? You're paying $100 for approximately $4 in expected value.

The Opposite: Undervalued Packs

Not all divergence is bad. Some packs show negative Claimed EV but positive Adjusted EV. These are packs the platform undervalues:

Pack	Platform	Claimed	Adjusted	Why
Platinum TCG	Beezie	-7.3%	+914.4%	Cards appreciated massively on secondary market
Wildcard	Beezie	-5.6%	+212.5%	Market value far exceeds platform pricing
Silver TCG	Beezie	-7.5%	+71.7%	Steady card appreciation since listing
Gold TCG	Beezie	-6.0%	+11.5%	Moderate undervaluation

Why does this happen? Beezie packs use median_7d calibration, not pool-weighted. These packs don't have finite prize pools — every buyer draws from the same odds. But the cards inside have appreciated significantly on secondary markets since the pack was listed. The platform's Claimed EV still uses the original card valuations.

Platinum TCG at +914% seems absurd. Verify the liquidity before committing. A pack can have astronomical Adjusted EV but if the cards inside have thin order books, you can't realize that value.

What Calibration Confidence Actually Means

Confidence	Sample Size	Meaning
High	200+	Reliable — large enough sample to trust the calibration
Medium	50-199	Directionally correct but could shift with more data
Low	<50	Estimate only — treat as a signal, not a fact

Right now, 45 out of 46 tracked packs have calibration data, with most Phygitals packs at high confidence. Beezie and Renaiss packs also show high confidence using the median_7d method.

Using EV Divergence as an Early Warning

The gap between Claimed and Adjusted EV isn't just a static number — it's a trend indicator.

When you see a pack where:

• Claimed EV is stable (the platform doesn't change its math)
• Adjusted EV is declining week over week
• Pool depletion is crossing 50%

That's your warning signal. The pack is transitioning from genuinely profitable to a trap. The 50% pool depletion threshold is where the divergence typically accelerates — once half the high-value cards are gone, the remaining EV drops exponentially, not linearly.

The Divergence Acceleration Pattern

Pool Depleted	Typical Claimed vs Adjusted Gap	Signal
<20%	Within 5%	Safe — odds still roughly as advertised
20-50%	10-30%	Caution — verify before buying
50-80%	30-80%	Danger — most high-value cards gone
>80%	80-96%	Avoid — you're buying empty lottery tickets

Decision Framework: When to Buy, Watch, or Avoid

Combine all the calibration signals into one decision:

Right now, 5 Phygitals packs are in AVOID territory and 4 Beezie packs are in WATCH/BUY territory. Check GapSense.uk/app for the latest calibrated numbers — they update every 30 minutes.