Portfolio review · NFAI engine
The engine evaluates once a day at the close: stops, exposure, ratchets, concentration, and the circuit breaker against SPY. No intraday decisions.
Engine
Running
Last update
July 31, 2026
Source
Live feed
Signals
11
Breakers
1
The daily optimizer · what would I build today?
DefensiveDefensive is the posture today's gates support, and the book is one the engine would rebuild from cash. Top move: Raise cash — the circuit breaker is tripped.
The optimizer ranks today's ACTIONS from the same gates the review reads — it invents no forecast and recommends in words, never trades. Its own calls are not yet graded, so it sits on the Investment Committee with its trust withheld until reality has marked enough of them.
If the engine started from 100% cash today
DefensiveStarting from 100% cash today, the engine would rebuild this book: all 12 names it would open again — 4 at full size, 8 smaller. It would stay defensive and keep powder dry: the circuit breaker is tripped; book risk reads High.
Since start
−5.6%
SPY
+8.2%
Relative
−13.8%
Circuit breaker active
Portfolio DD −8.2% · SPY DD −1.6% · Relative −6.6%
Action: reduce speculative exposure.
Top risk: KGC 16.1% · RDDT 14.7% · HALO 13.6%
Within limits
Since start
+2.6%
SPY
+8.2%
Relative
−5.7%
Within risk limits
Portfolio DD −3.7% · SPY DD −1.6% · Relative −2.1%
Top risk: HALO 40.6% · NTES 26.3% · WM 13.5%
Risk concentration: HALO 40.6% over the risk limit.
Within limits
Conviction is how strongly each rule fired, not a probability of profit.
The engine's read for the experiment, not financial advice. Stops, exposure, and ratchets are computed from price. Buy scoring requires a fundamentals feed that is not wired in yet, so the engine does not issue buys.
A re-underwrite of the existing book plus cash — it cannot discover names it does not hold, it composes reads graded elsewhere (thesis, stops, credit, risk, regime), and it recommends in words, never trades. The gap between what is held and what would be chosen is the endowment bias, measured.
Conviction is how strongly each rule fired, not a probability of profit.
It is close to its stop; if it breaks, the position gets sold.
Last $35.67 · Stop $34.49 · 3.3% above
Proximity to stop 0.34 · Position weakness 0.89
It is close to its stop; if it breaks, the position gets sold.
Last $140.67 · Stop $139.80 · 0.6% above
Proximity to stop 0.88 · Position weakness 1.00
It is close to its stop; if it breaks, the position gets sold.
Last $23.10 · Stop $22.17 · 4.0% above
Proximity to stop 0.19 · Position weakness 0.84
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 14.7% · Target 12.0% · Reduce 2.7%
Over size cap 0.22 · Risk share 0.45 · Correlation to book 0.29
It is running, so the stop trails the high and lets the gain ride.
Up +41.2% · trailing stop engaged
Gain past trail trigger 0.04
It is close to its stop; if it breaks, the position gets sold.
Last $23.10 · Stop $22.28 · 3.5% above
Proximity to stop 0.29 · Position weakness 0.88
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 20.4% · Target 12.0% · Reduce 8.4%
Over size cap 0.70 · Risk share 0.88 · Correlation to book 0.34
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 23.1% · Target 12.0% · Reduce 11.1%
Over size cap 0.92 · Risk share 1.00 · Correlation to book 0.29
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 18.4% · Target 12.0% · Reduce 6.4%
Over size cap 0.53 · Risk share 0.45 · Correlation to book 0.48
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 13.7% · Target 12.0% · Reduce 1.7%
Over size cap 0.14 · Risk share 0.00 · Correlation to book 0.43
It has grown past its target size, so trimming keeps one name from dominating the book.
Exposure 17.4% · Target 12.0% · Reduce 5.4%
Over size cap 0.45 · Risk share 0.40 · Correlation to book 0.27
