Disapointing June 2026 filter performance, weighted down by pre-Agentic review VNA_Expected Return
If purchased at the closing price of 6/12/26, the average of the 6 tickers in our 6/11/2026 close-based filter returned -3.45% through the 7/13/2026 close. This lagged the average price return of the 129 tickers in the screened universe by 171bps (they averaged -1.74%) and lagged the SPY by even moreson, which came in at -1.00%
Here is the updated average monthly performance of the filter tickers against the SPY and the trade universe average since we started tracking it in December of last year. Note that the filter output is always presumed to be purchased at the closing price of the day following its date of being published.
See below for a comparison of the distribution of the returns for the tickers that met the 5/11/26 filter criteria (orange) and the population they were drawn from (blue), through 6/10/26. Filter winners included BIIB and MRK. Filter laggards included FCX, TEVA,POST, OXY.
So what drove the underperformance we saw last month?
The VNA Target Price, pre-agentic VecEvent review of specific, more recent than LLM context type sources, proved to be a very contrary indicator last month. This is only one month of production history, and we have no back-test to draw reassurance from, given the dependence on LLM’s. It could be that we need to give it more time to play out. Or it could be that the LLM baseline VNA_PctUpside, which depends entirely on LLM sourced VecEvents, is not sufficiently current. Afterall, the training data it incorporates is typically 3-6 months old, and our economy and markets are very dynamic. For example, if you look back at our results from last month you will see that an agentic review of the VecEvents using only the most recent earnings call transcript resulted in some fairly material price revisions, half of which were to the downside. A filter using agentically reviewed VecEvents may have come up with very different results.
We continue to believe that with the right VecEvents and associated characterizations the VNA_PctUpside metric will prove to be a powerful indicator of forward price performance. But for now, we view the baseline (pre-Agentic review of VecEvents incorporating recent information) VNA_PctUpside with some suspicion.
What does the filter look like this month?
We pause targeting its ostensibly bullish end this month, and instead target the middle of its distribution (which performed well, though not as well as the bottom of the distribution) within the subset of tickers that also meets our long running and still effective EDB, EUB and V-Score criteria (> -6%, >4%, >=0, respectively).
The tickers that meet our VNA-neutral criteria for mid July – mid August are ORLY, AMZN, COST, IRM, PHM, and QQQ. We will continue to adjust the filter as we gather more performance data on the VNA_PctUpside metric.
