diff --git a/docs/FAST_MVP_PLAN.md b/docs/FAST_MVP_PLAN.md
index 2f3f2b4..4a73405 100644
--- a/docs/FAST_MVP_PLAN.md
+++ b/docs/FAST_MVP_PLAN.md
@@ -804,6 +804,61 @@ Static test data can hide critical bugs like stale caches.
 
 ---
 
+## 🔍 UPDATE: Reserve Refresh Already Implemented (Nov 30, 2024)
+
+Upon inspection, **reserve refresh was already implemented** in:
+- `pkg/discovery/reserve_refresh.go` - V2 pools
+- `pkg/discovery/reserve_refresh_v3.go` - V3 pools
+- `cmd/mev-flashloan/main.go` lines 141-149 - Called before each scan
+
+**The REAL issues are:**
+
+### 1. Limited Pool Coverage
+- Only 9 UniswapV2 pools discovered
+- No UniswapV3 pools in cache (only parsing, not discovering)
+- Need 100+ pools for meaningful arbitrage opportunities
+
+### 2. Same-Protocol Only
+- Current detector only checks V2→V2 arbitrage
+- REAL opportunities are V2→V3 cross-protocol
+- Same-protocol markets are too efficient
+
+### 3. Market Efficiency
+- Professional MEV bots with:
+  - Flashbots/private mempools
+  - Sequencer-level access
+  - Submillisecond execution
+- Public RPC bot can't compete on same-protocol arb
+
+### Required Fixes (Priority Order):
+
+| Fix | Impact | Effort |
+|-----|--------|--------|
+| Add UniswapV3 pool discovery | HIGH | 2 hours |
+| Enable cross-protocol arbitrage (V2↔V3) | HIGH | 4 hours |
+| Add SushiSwap, Camelot pools | MEDIUM | 3 hours |
+| Reduce scan interval to 5s | LOW | 5 min |
+| Add sequencer feed integration | HIGH | 8 hours |
+
+### Why Cross-Protocol Matters
+
+```
+Same-protocol (V2→V2):
+  Pool A: 1 WETH = 3000 USDC
+  Pool B: 1 WETH = 3001 USDC
+  Difference: 0.03% - Too small, eaten by gas
+
+Cross-protocol (V2→V3):
+  V2 Pool: 1 WETH = 3000 USDC (constant product)
+  V3 Pool: 1 WETH = 3050 USDC (concentrated liquidity)
+  Difference: 1.67% - PROFITABLE!
+```
+
+V3 pools have different pricing due to concentrated liquidity,
+creating larger arbitrage windows with V2 pools.
+
+---
+
 ## 📋 Updated Implementation Checklist
 
 ### Immediate Fixes Required: