NeuroCycle
- 指标
- 版本: 2.82
- 激活: 5
NeuroCycle is a self-learning indicator designed primarily for binary options and short-term scalping.
The system automatically searches for suitable parameters of its built-in trading model for a selected market period.
The main optimization principle is what I call the Rule of Three Variables: Time, Sample Size, and Probability.
The idea is simple:
Time
The training period should be long enough for the market to pass through different conditions and regimes.
Sample Size
A sufficiently large number of signals must be collected during this period, so that the result is not based on just a few lucky trades.
Probability
The resulting sample must demonstrate a sufficiently high percentage of successful signals.
TRAINING
In practice, I have found that NeuroCycle should preferably be trained on at least 6 months of historical data.
The training process itself usually takes only a few minutes, depending on the computer and the amount of available history.
With the default settings, the indicator requires at least 100 signals per month within the training period. This prevents the algorithm from selecting an attractive win rate based on an excessively small sample.
As a result, we obtain a set of parameters that demonstrated stable performance throughout the entire selected training period.
MARKET INERTIA
The most interesting part begins after the training period ends.
Market patterns usually do not disappear instantly. Parameters found on a recent historical period may continue to work for some time after that period has ended.
In live market conditions, the actual success rate will usually be somewhat lower than the TRAIN statistics. This is natural: the future never perfectly repeats the training period.
However, with good-quality training, the probability margin may still be sufficient to maintain positive statistics.
From my observations, this kind of market inertia may persist for several months. However, I recommend retraining the indicator every month.
A typical workflow:
6 months of training -> 1 month of trading -> retraining
Example:
TRAIN: 01.01.2026 - 30.06.2026
Trading period: 01.07.2026 - 31.07.2026
After that, the training range is shifted forward and the model is retrained using more recent market data.
A longer training period may make the discovered patterns more stable, but a longer period does not automatically mean a higher win rate.
The important part is maintaining a balance between all three variables: Time, Sample Size, and Probability.
BOLLINGER POST-FILTER
NeuroCycle includes an additional Bollinger post-filter.
This filter does not participate in model training. Instead, it filters the signals generated by the trained model.
Its purpose is to reduce the number of signals appearing inside market noise and retain more signals near local price extremes.
This is a deliberate trade-off:
Fewer signals -> potentially higher signal quality.
One of the most important filter parameters is Deviations.
The default value is 2.0, which is a good starting point in most cases.
If you use NeuroCycle on many currency pairs and do not have a shortage of signals, you can experiment with values of 2.5-3.0.
This restricts the area in which signals can appear and usually reduces their number significantly.
HOW TO USE NEUROCYCLE
1. Attach the indicator to an M5 chart.
Select a training period of at least 6 months. You may experiment with a longer historical depth if desired.
The main working timeframe of NeuroCycle is M5.
This makes the indicator particularly suitable for short-term trading and 5-minute binary options.
The algorithm can also work on M15, but the number of signals decreases considerably, which results in a weaker statistical sample.
2. Wait for training to complete.
During training, a progress indicator is displayed in the upper-left corner of the chart.
Once training is completed, NeuroCycle automatically displays the discovered signals and their performance statistics.
3. Check the TRAIN WR.
A colored training-quality indicator appears in the upper-right corner.
It shows the percentage of successful signals during the TRAIN period.
As a general guideline:
Green: good training quality
Yellow: borderline result
Red: not recommended for trading
For practical trading, I primarily look for a TRAIN WR of 55% or higher, provided that the sample size is sufficient and there are at least 100 signals in every month of the training period.
A very high win rate based on a small number of signals is considerably less interesting than a slightly lower win rate obtained from a large and stable sample.
4. If the message NO VALID PROFILE appears, it is usually better to skip that pair.
This means that the algorithm was unable to find a parameter set within the selected training range that simultaneously satisfies the required sample size and signal-quality conditions.
There is no need to force NeuroCycle to trade every currency pair.
Rejecting unsuitable market conditions is just as important to the system as finding suitable ones.
SIGNALS
NeuroCycle operates using a non-repaint principle.
Signals are generated without using future data.
Once a signal appears, it is not moved to another candle and does not disappear from history.
Historical statistics are calculated according to the same rules used to generate signals after the training period.
The indicator can also automatically compare several expiration periods and select the most suitable one for the model it has found.
For binary options, the primary focus is short expiration trading on M5, including 5-minute expiration setups.
ALERTS
For convenient monitoring, NeuroCycle includes:
- Sound notifications for new signals
- Optional visual notifications
- Display of recent signals from other currency pairs when NeuroCycle is running simultaneously on multiple charts
This makes it possible to monitor an entire portfolio of instruments without constantly switching between MetaTrader charts.
CONNECTING TRADING ROBOTS
The indicator provides open signal buffers for connecting Expert Advisors and external trading systems:
Buffer 0: BUY
Buffer 1: SELL
Therefore, NeuroCycle can be used either as a standalone arrow indicator or as a signal source for automated trading systems.
IMPORTANT
NeuroCycle is a statistical tool.
A strong result on historical training data does not guarantee the same result in future market conditions.
Markets constantly change, so I recommend regularly retraining the model, using a sufficiently large historical sample, and evaluating performance across multiple periods.
The main principle behind NeuroCycle is simple:
Do not search for a perfect setup based on a handful of beautiful trades.
Search for a stable probability across a large sample and a sufficiently long period of time.
