The P/B multiple has climbed from 7.87x in 2024Q2 to 11.15x in 2026Q3 even as ROE fell from a 12.1% peak to 4.8%, a divergence that implies the market is pricing fee-franchise optionality and crypto-cycle beta rather than demonstrated earning power on equity. For a bank whose equity base is only 0.17 of assets, the 9.49x book multiple leaves little room for the reserve build or funding-cost pressure already visible in the income statement; investors should monitor whether tangible book growth of roughly 35% over ten quarters ($7.63 to $10.33) can be sustained without further equity issuance, and the 0.18 PEG appears to rest on growth estimates that the recent fee-pool contraction does not corroborate.

The DuPont structure here is unusual for a bank: asset utilization is modest (0.4%-2.6% ROA range) and NIM contributed essentially nothing in 2026Q2 (-0.0%), so the entire return profile rides on fee capture — trading, subscriptions, and crypto — which the data show contracted from $872M in 2025Q4 to $415M in 2026Q3. Because ROA has fallen faster than leverage could offset it, ROE now sits below its 2024Q3-2024Q4 trajectory; the 178.1% efficiency ratio in 2026Q3 confirms that costs are absorbing more than the entire revenue base, which suggests reported profitability is being supported by non-operating items rather than core franchise economics, consistent with the earlier finding of net income printed against operating losses.

The NIM collapse to near zero implies the customer credit balance franchise — essentially free funding — no longer earns a spread over the yield deployed on the $58.3B asset base, which is a structural concern rather than a typical rate-cycle effect given that customer balances are typically a low-cost liability source. Simultaneously, the efficiency ratio's swing from 42.3% in 2026Q1 to 149.6% and then 178.1% indicates the expense base is largely fixed while revenue halved, producing severe negative operating leverage; if the 2026Q3 NIM print (unavailable) confirms the 2026Q2 level, investors should assume the interest line has become a drag rather than a complement to fees.

A 0.17 equity-to-assets ratio leaves roughly 1.7x of reported book capital against the $301M peak quarterly provision level seen in 2025Q4, which suggests reserve capacity exists but is not generous relative to a balance sheet whose securities portfolio reached $10.3B against $8.6B of equity in 2025Q3. Because the capital decline appears liability-led rather than buyback-led — repurchases totaled only about $1.14B over ten quarters versus the asset expansion — the binding constraint on growth is the pace of customer fundings, and the $12B cash drawdown between 2026Q2 and 2026Q3 warrants monitoring as a potential indicator of how quickly liquidity, and with it the capital buffer, can move.