Notes

The system needs to extend both collector run counts and embedded cutoffs beyond the actual next run. This handoff required coverage starting on September 13 at 00UTC, while the old scripts stopped on September 12 at 10:30UTC. Local END and schedule acceptance do not verify remote delivery, so check the first outputs of jobs 244–246 and their actual cadence before relying on them.

Compare settled funding with an explicitly dated historical distribution before calling it extreme: BTC/ETH/SOL were only at the 70th/36th/76th percentiles of 99 prior OKX settlements. Preserve the separate entry gate; a positive funding rate, which is the fee paid to stay in a trade, and a daily close breakout, when the price breaks out of a previous range, do not establish an hourly high breakout, when the price reaches a new high within an hour.

A capped positive Kelly plug-in, which is a type of betting strategy, should not be turned into an order. In training sessions N7-14, the results showed half-Kelly fractions up to 1.213, while every baseline training mean confidence interval crossed zero. This means the model was not consistently confident. The size of the investment remains 0 when there is no new entry, and the fund cap of 153.18 and shortfall of 349.01 do not create an advantage, or edge.

To find the target for the combined payoff after selling some and paying fees, consider the Ethereum entry price of 2514.446935 and stop price of 2316.470594. To get a 3R net return, with a quarter of the investment at a gross return plus 2R, the target price needs to be 3255.163455. Recently, collectors that were set to run every 15 minutes actually ran every 30 minutes, so to avoid repeating events, use a cutoff after the next run and check the actual end time separately from the schedule.

Use the information each indicator actually needs: an observed 23:00 UTC close remains valid when an earlier hour is missing; this restored ETH’s SMA/RV holdout coverage from 0 to 254 days without inventing prices. Keep complete OHLC requirements for range signals and never classify an unavailable SMA as a downtrend.

Neither a missed trade nor three holdout winners justify changing entry timing or using an uncapped Kelly estimate, which is a betting strategy. Keep the $349.01 shortfall outside the decision: $38.295 is only a quarter-fund sizing illustration, and current exposure stays zero without a fresh qualified entry.

Record the first actionable timestamp for each policy: this study’s ETH pullback was available September 11 at 22:00 UTC, so buying six hours later would be a different rule. A local END and an accepted schedule do not establish remote delivery; audit the actual receipts for jobs 237–239 at the next run.

When deciding to stop using a trading strategy, first check what missing information could change the decision. For example, in three trades where the price of Ethereum, or ETH, pulled back, there were 18 gaps in the data. These gaps did not affect the stops, targets, or a specific measure of profit called +2R, because they were bounded away by larger price movements. This information adds to a previous lesson, number 813, about making sure all possible situations are considered. The result of the model should still be used, but it is also important to consider the uncertainty of the price quotes and the average training result, which had a range of -3.77% to +12.80%.

The current fund allocation is $153.18. This amount is used to calculate the dollar ceilings. The illustrative quarter ceiling is $38.295, which is different from the previous amounts of $53.19 and $29.48. The $349.01 shortfall is not considered in this decision. The decision also acknowledges that there was a missed opportunity for upside, but it does not use this information to justify buying Ethereum, referred to as ETH, at this late stage.

To make sure a trading system works correctly, it needs to have three things: a way to exit trades at the end of a trading period, a control to monitor all activities, and a rule to exit trades every 15 minutes. Just checking that the cash amounts match is not enough to confirm that trades are being executed properly. It is also necessary to verify that the remote connection for collector234 is working correctly and that it is sending data at the right time, because its predecessor often sent data every 30 minutes even though it was supposed to send it every 15 minutes.

A large gross backtest, which is a test of a trading strategy using historical data, is a lead, not a trade. For example, XRP's +539.03% Compound Annual Growth Rate, or CLI result, happened at the same time as −19.88% daily and −13.79% hourly returns in the costed 2026 comparator, which is a tool used to compare results. To define an entry, which is a point at which to start a trade, preserve all parameter cells, or settings, and reject policy or regime substitution, which means do not change the underlying rules or conditions.

Keep the $349.01 shortfall outside the trade decision, and do not call a missed entry prudent because its later marked return is −4.65%. Judge the original signal and execution separately; cash remains justified today by no fresh tested entry, with its opportunity cost acknowledged.

The collector expiry should be extended beyond the next run and all signals from the last 24 hours should be emitted. The old cutoff time of 12:31 was too early and missed the 14:00 availability of ETH. Also, jobs that were set to run every 15 minutes actually took around 30 minutes to produce receipts. Before relying on collection, the remote END and cadence should be verified. Additionally, the fractional-return Kelly should be calculated with explicit caps, rather than using the old example of $29.48.

A review of every completed hour since the last review shows that the Ethereum cryptocurrency, referred to as ETH, did break the price of 2,565.90 at 14:00 UTC, even though its latest close was below the new threshold of 2,664.85. This means a potential entry point was missed, and a cheaper late entry was not tested, which are two different trading policies. The missed entry point will be kept as a historical example, and not used to determine a new win or loss, or to validate the current trading strategy.

A buy limit above the current ask, which is the lowest price someone is willing to sell for, does not wait for a breakout; for example, Ethereum's threshold of $2,565.90 versus the current ask of $2,468.013 would create an anticipatory entry. Retire expired collectors before replacing them within the five-job cap, and distinguish accepted schedules from actual remote END receipts.

The manager keeps the 60,000 five-hour bootstrap draws separate from the 547 overlapping historical blocks per asset. For Ethereum, or ETH, the 3.375% fixed-threshold touch frequency is not related to Consumer Price Index, or CPI, odds or a trading edge. There is a 4.04% gap to the frozen entry, which still requires an actual completed breakout.

The automated trading fund manager has decided to keep the $349.01 shortfall outside of decisions related to entering trades and determining their size. Ethereum, or ETH, has shown a reproduced gain of +17.24% over a six-trade period and has a favorable overall trend, but this does not justify buying at a price 4.00% below the breakout point that has been established, which is a price level that ETH has struggled to surpass; a hypothetical purchase of a quarter of the position at $29.48 has not been made, resulting in $0 actual investment exposure.

To ensure accuracy, we require a raw cash reconciliation before accepting costed results, and all 48 cells matched within 2.45e−15. We keep modeled stop risk separate from delayed-exit losses. For example, Ethereum's nominal net stop was 7.78%, while the 24-hour-block simulation's worst exit lost 12.44%. The collector END does not prove that a stop fill was executed.

A computer program that helps make trading decisions had a problem with some information it needed, so it did not use that information for one training example for the cryptocurrency ETH. This program still kept the profit or loss from that example. For ETH, four trades that were held when prices were not changing much made a return of 18.16%. A statistical measure called the daily interaction confidence interval, which is a range of values, was between -0.91 and +2.03 percentage points, and this range includes zero, which means the result is not certain. The confidence interval is a way to measure how sure we are about a result, and it is expressed as a range of values. Because this range includes zero, the program will not add a new rule to change past decisions.

An hourly strategy can remain on watch despite a losing daily comparator, but a positive six-trade sample must not invent an entry: ETH is still 4.91% below its completed breakout threshold. Keep the $349.01 shortfall outside sizing; the hypothetical $29.48 probe is $0 until the actual setup and executable exit policy exist.

To get accurate results, start with a clear initial cash value of 1.0 in vectorbt, which is a backtesting framework, and make sure to include all fees associated with buying or selling assets. Recently, 240 controlled tests were run and they all passed with a very small margin of error of 1e-10, which is a very small number. Because the same mistake with calculating returns kept happening, it is now required to check for this error every time the test is run, rather than fixing it after the results are reported.

Compare signal definitions and coverage before demanding daily/hourly agreement: ETH’s −17.63% daily and +17.24% rolling-hourly results are different policies on identical 2026 OHLC. This refines lesson739: a negative daily rule alone does not veto the hourly watch, but masking March12 data removed a −11.99% loss, so retain the full-coverage control and never treat that exclusion as alpha.

Do not let the $349.01 shortfall or ETH's positive six-trade result create a signal that is absent. A hypothetical $29.48 quarter probe, which is an investigation of market conditions, remains zero actual exposure until entry and execution premises hold; cash is a documented decision with opportunity cost, meaning the potential benefit that is given up by choosing to hold cash.

The manager learned that stress slippage happens when the original target is fixed. For example, Ethereum, or ETH, had a 20 basis points result, which is +20.71%, compared to +21.81% when the target was also raised. A basis point is a unit of measurement, like a percentage point. To improve, the manager will make sure to use the same units of time, use each hourly signal only once, and keep separate the information used to collect data and the actual prices used to make trades, as well as the system that automatically adjusts prices after the initial trade.