The Pharos desk · paper · simulated fills
When it is on, my terminal runs a paper book that trades what Pharos calls, on E-mini NQ (NQ, $20 a point), one contract, one position at a time. Right now it fades last night’s Globex high and low from 09:40 ET: a limit 2 points in front, target 50, stop 120 through the level, 1 touch a day; the high and low of the first half hour from 10:00 ET: a limit 2 points in front, target 15, stop 100 through the level, 1 touch a day. It stops for the day after 100 points of losses and is flat by 15:55 ET. Every fill is simulated on the live tape and lands in the terminal’s own journal with a note. Every trade since the record began (Oct 1, 10:20 ET) is on this page, ten minutes behind the tape like the picture, misses included, with 0.5 pt per round trip charged on every trade.
The desk is off while I rework its rules: nothing trades until it is armed again, and this page says so the moment it is. This record runs since Oct 1, 10:20 ET, so these are the numbers of a very short record and not a verdict on anything. Before the desk moved to NQ it traded ES from 2026-09-14 to 2026-10-01: 62 trades, 79% wins, -21.20 points after costs.
SIM · disarmed
No closed trade on the record yet.
The ES desk taught me something first. Its rules were picked as the best cells of a large grid on the same months of tape they were then scored on, and on live days they earned close to nothing: +7 points over 60 trades before costs, negative after them. When I replayed those live days through the same simulator, it matched the record trade for trade, so the fills were fine. The rules simply had not been tested on data they had not seen.
So for NQ the test was out of sample. NQ moves about seven times as many points as ES in every session, so every distance was sized to NQ and measured, not assumed. I tuned on May to July 2026 and judged on August and September, with two points of slippage charged on every stop (what the NQ tape actually costs a stop order within half a second) and half a point on every round trip. Most of the ES rules failed that test on NQ: the Asia-session fades, yesterday’s close, the walls and the gamma flip all lost money on the months they were not tuned on. What held was fading last night’s Globex high and low in the cash session: 31 trades on the held-out months, 77% wins, about +17 points a trade after costs, and every neighbouring setting positive too. Fading the first half hour’s range held as well, more often and for less per trade.
Read the cost of a loss before the win rate. With a 120-point stop one losing trade is about $2,500 on one NQ, and in the held-out months roughly one trading day in four ended red. The second half of September was weak, the same weeks the ES rules failed in. That is why the desk is on paper, with the record kept in the open. The rules change when the record says so, not before.
Hypothetical performance results have many inherent limitations. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown; simulated results may under- or over-compensate for slippage, commissions and the impact of the trades themselves. Nothing on this page is advice.
Today’s walls, free →·The walls’ graded record →·The terminal, five days for $0 →
