Skip to content
AlgoVerdict
Menu
Allylla Pullback Bot product shot

EA review ·

Allylla Pullback Bot

provisional

Allylla Pullback Bot

5 signals on real-money accounts (per MQL5 rules), selected by the vendor

The vendor lists 5 public signals. We checked that they exist and can be viewed. The seller still chooses what to show, so such evidence earns at most 8.0 for verifiability.

Checked
Signal exists and is public · real money per the MQL5 signal rules, broker’s live server · trading behaviour matches the described design
Not shown
Return, drawdown, running time. Past figures tempt buyers more than they inform them.
Counts towards
Verifiability (25% of the grade): 5.2 out of 10
Strength of evidence
2 of 3: selected by the seller
IntradayXAUUSDMT5

Reviewed by Pitt Petruschke, operator. No stake in Allylla Pullback Bot; receives commission from ATS and develops his own EA, Dawnbreak. Disclosure

Cross-checked by an AI agent (different model, blind): 6 of 6 criteria agree. Differences

Verdict

Intraday EA on gold, stop on every trade. Evidence chosen by the seller only, weakest point verifiability: 6.4 out of 10.

Suits you if …

  • Yes: you want an EA with fixed entries and exits
  • Yes: you run it as one building block in a portfolio
  • Caution: you set setfile and risk yourself, without ATS calibration

Evidence behind this verdict

  • MQL5 product page Allylla Pullback Bot (price $299, rent $149/month, version 2.80 of 12 Nov 2026, published 11 Nov 2026, strategy, requirements, recommended risk, 5 reviews)read 17 November 2026
  • MQL5 product page Allylla Pullback Bot, "What's new" (version 2.80)read 17 November 2026
  • MQL5 product page Allylla Pullback Bot, comments (incl. #6 by the developer on the signals' default settings, 17 Nov 2026)read 17 November 2026

Design & risk

Allylla Pullback Bot by Uttam Kumar Nandeibam is a Intraday EA on XAUUSD (MT5), sold on the MQL5 market. Each position has its own exit; losses are not absorbed by averaging in. What matters is whether execution at your broker matches the described process. You buy the vendor default; risk frame and setfile are up to you.

Newsletter · Sundays

New EA reviews, by email on Sundays

Before you buy your next EA: once a week, the digest brings the new and updated reviews.

  • New EA reviews with grade and verdict
  • What changed in existing verdicts
  • A note when something material changes on a signal we track

By subscribing you agree to receive the digest by email. You confirm via a link (double opt-in) and can unsubscribe anytime via the link in every email. Details in the privacy policy.

The full review

Overview

The Allylla Pullback Bot by Uttam Kumar Nandeibam is a gold EA for MetaTrader 5 that, per the vendor, trades retracements after structural breaks: one entry per strategy, a stop-loss, no grid, no martingale. In its favour are this design per the vendor, openly stated loss limits and five public signals at four brokers. Against it stand a vendor-recommended risk of 3 to 5% per trade, very high leverage on the signals and a track of a few weeks whose growth mostly came before the market launch. Our grade is provisional.

What the vendor sells

Per the product page and setup guide (read 17 Nov 2026):

  • Strategy: Per the vendor, the EA looks for structural breaks across several timeframes and then waits for retracements to certain pullback levels. The entries come from an in-house price-action logic per the description, not from indicators and not from pending orders at breakout levels. Eight entry models are grouped into five strategies that can be switched off one by one.
  • Stops: Per the vendor every entry gets a stop-loss. The loss limit per strategy is between $35 and $75 per 0.01 lot and can be adjusted. Per the setup guide the stop can sit on the broker's server or be watched only locally in the terminal; in the second case the terminal has to run all the time. On top come time-based exits that can close a position at breakeven or with a small loss after a set time.
  • No grid, no martingale: The vendor stresses that the EA does not add to losing trades and has no hard-coded blocked dates that flatter backtests.
  • Simultaneous trades: By default each of the five strategies may hold its own position, so positions can overlap.
  • Account: hedging account recommended; the vendor names $1,000 as the ideal account size for a fixed lot of 0.01. Defaults are a fixed 0.01 lot and a news filter for major US events.
  • Price: The product page says: "Offer Price: Get it today @ $299 (only 3 copies left of 15 at this discounted price) | Regular Placement Price: $599" (read 17 Nov 2026). It also announces: "the price of this EA will increase in steps of $50 / $100 for every certain number of copies sold" (read 17 Nov 2026). Whether and when the price rises we will check at the next update. Rent $149 for one month, free demo, ten activations. Published 11 Nov 2026, current version 2.80 of 12 Nov 2026 (per the version history only a bug fix with no change to the trading logic).

The setup guide with short notes on every setting is public on the vendor's MQL5 blog. That is more than many vendors show before the sale.

The recommended risk

Under "Important Note" the vendor writes: "The suggested risk per trade is around 3% to 5% of equity." (product page, read 17 Nov 2026). The same note says the signals are for demonstration only and their risk settings may not suit every account size.

The recommended risk is high. At 3 to 5% per trade, a run of a few stops costs a large part of the account, and because up to five strategies can be open at once, the risk adds up in a bad phase. The vendor himself writes that drawdowns and dry spells can drag on when the market does not suit the strategies. The built-in loss limits go even further: up to $75 per 0.01 lot is up to 7.5% per trade on the recommended account of $1,000 per 0.01 lot, more than the vendor's own recommendation of 3 to 5%. If all five strategies hit their stops at once, that is up to 37.5%, over a third of the account. On accounts much smaller than recommended, a single stop can cost most of it. Anyone testing the EA should set the risk well below what the product page suggests and lower the loss limits accordingly. How to derive the lot size from risk and stop is in the guide Position sizing and risk per trade.

The five signals

The vendor links five public MQL5 signals at four brokers:

SignalVendor's labelBrokerstarted
2397645 "Balanced Exposure"balancedRoboForex29 Oct 2026
2396795 "High Exposure"high exposureRoboForex23 Oct 2026
2398788 "ALLYLLA Pullback Bot 100"high exposureIC Markets6 Nov 2026
2398859 "Allylla High Exposure 150"high exposureVantage7 Nov 2026
2398791 "ALLYLLA High Exposure 200"high exposureTMGM6 Nov 2026

The MQL5 signal rules only allow real-money accounts, and all five pages mark the account as a Real Account on the broker's live server. Per a comment by the developer of 17 Nov 2026 they run the default settings. None had subscribers; copying costs $999 a month each. What we read on 17 Nov 2026, without figures, as in all our reviews:

  • Very short run: The signals have run for a few weeks. All five started before the market launch on 11 Nov 2026, three of them only a few days before. MQL5 warns on all five that the account is newly opened and the results may be random, and that the number of deals is too small to judge.
  • Growth before the launch: Per the history data on the signal pages, most of the growth on all five accounts came before 11 Nov 2026. On the three newest accounts more than half even dates from before the official signal start; MQL5 shows these trades before the start marker. Since the launch all accounts have grown much more slowly.
  • Marked decline, measured by equity: By balance, i.e. by closed trades, the decline is modest on all five accounts. That says little, because almost all trades closed in profit. What counts is the worst value from balance and equity: the "High Exposure" account (2396795) was at times well underwater. So trades at times ran far against the account before closing in profit. This value counts for the grade; it falls into a middle band of the table in the calculation.
  • Hardly any triggered stops: There have been hardly any losing trades so far, on most accounts none at all. What triggered stops do at 3 to 5% risk per trade is therefore not yet in the track.
  • Same trades, different account sizes: The two accounts "Balanced Exposure" and "High Exposure" at the same broker show largely the same trades with the same dollar amounts. The "High Exposure" account, though, has only a fraction of the capital, so the same floating loss weighs many times more there. The three other accounts also run with far less than the recommended $1,000 per 0.01 lot. With a fixed 0.01 lot and a loss limit of up to $75 per trade, a single stop can cost most of such an account.
  • Very high leverage: All five accounts run with very high leverage, one with an extreme one. On the "High Exposure" account (2396795) a noticeable part of the capital was at times tied up as margin.

We do not show the figures themselves in the text; that applies to every EA on AlgoVerdict (methodology). How to read signal metrics and why we take the worst value for the grade is in the guide Trading metrics.

Leverage and margin at EU brokers

The signals trade with leverage that EU-regulated brokers do not offer retail clients. There the cap on gold is 1:20. Each position then ties up a multiple of the margin the signal accounts need for it. As long as a single 0.01-lot position is open, that is no problem on an account of the recommended size. But when several of the five strategies run at once, margin on small accounts gets tight fast, and rejected orders or an early margin call are possible. Anyone wanting to mirror the signals at EU leverage cannot copy the position sizes of the "High Exposure" accounts.

What buyers report

On the MQL5 market the EA stands at 5 of 5 from five reviews, plus eleven comments (as of 17 Nov 2026). All reviews date from the first week after launch. Buyers praise the easy setup with default settings, the patient trading style and quick answers from the developer, including in his Telegram channel. Reviews after a few days say little about an EA's risk; they do not count towards our grade.

In the comments a buyer reports that the EA had shown the status "history loading" for hours and was not trading. The developer replied that this was not normal with the market open and asked for the log files. How it ended is not visible on the page.

The seller's overall rating across all his products and signals is 4 of 5 from 30 votes (as of 17 Nov 2026). The vendor also sells two other gold EAs, Osloma Gold and Osloma FlipPro MT5, which use grid management per their product pages. Per the vendor, Allylla is deliberately built differently.

How the provisional grade comes about

We grade the Allylla Pullback Bot on the same six criteria as every EA (calculation) and deliberately go lower where the data formula would be too kind:

  • Risk control: The starting point is the worst drawdown from balance and equity across all five signals, here the marked decline by equity on the "High Exposure" account. We deduct for the high recommended risk and the loss limits that already exceed it on the recommended account, and additionally for the very high leverage and the fact that the stop-loss has hardly been triggered in the short track.
  • Track record: Five public real-money signals count, the data formula would give more. We cut clearly: a few weeks, all started before the launch, most of the growth from before then, small accounts, chosen by the vendor.
  • Transparency: Design, loss limits and settings are described publicly, which counts. Deduction because the logic is only outlined, and because the vendor on the one hand says the signals run the defaults and on the other calls them "High Exposure" and writes that their risk settings may not suit every account.
  • Price and value: We grade the offer price of 17 Nov 2026. If it rises as announced, we re-grade.
  • Usability: neutral mark, we have not tested the installation ourselves.
  • Broker independence: four brokers on record, which is broad. Clear deduction because all accounts are run by the vendor with very high leverage and three of the four brokers have only been on board since early November, with hardly a triggered stop.

As with every EA, an AI from a different model family also grades the EA blind; where it differs strongly on a criterion, the mean counts until a human has re-checked. For the Allylla Pullback Bot the second review is close to our grade on all six criteria. We will re-check the grade once the signals have run for several months, the price rises or a larger decline occurs.

Who it does (not) suit

If you want to test it: first the free demo in the strategy tester across several periods, then a separate small account of the recommended size, risk well below the vendor's recommendation, the stop with the broker rather than only in the terminal, and only money you can afford to lose. Follow the signals for several months before adding capital. Because the stop can be watched in the terminal, the EA belongs on a VPS.

Not suitable for anyone who needs a proven track record, for buyers who extrapolate the curve so far, for small accounts at the recommended risk and as a second gold EA next to another gold system such as The Gold Reaper: all sit in the same instrument, which is the same bet several times (see EA portfolio management). How backtest and live can drift apart on gold is shown in Gold EA: backtest vs live. An overview of reviewed gold EAs is in Best gold EA 2026.

Verdict

Per the vendor, the Allylla Pullback Bot is a pullback EA for gold with one entry per strategy and a stop-loss, without grid and martingale, and the vendor discloses loss limits and settings more openly than many others. Against that stand a recommended risk of 3 to 5% per trade, loss limits that go even beyond it, a marked decline by equity on one signal account, very high leverage on the signals and a track of a few weeks whose growth mostly came before the launch and in which hardly a stop was triggered. Our grade is therefore provisional; we grade the EA with the same formula as every EA. We earn nothing from the EA itself; we link the vendor and MQL5 only as sources, never as partner links. Links marked “Ad” on this page (VPS, ATS-listed alternatives) are partner links. Past results are no guarantee of future ones, and trading leveraged products can lead to the loss of the capital invested.

What you need to run it

  • Broker with low spreads on XAUUSDRaw or ECN account, no EA restrictionsrequired
  • VPS close to the brokerThe EA has to run around the clockrequired
  • Fix the risk per tradeChoose a lot size that fits the accountset it up yourself

For and against

In favour

  • Per the vendor one entry per strategy with a stop-loss, no grid and no martingale
  • The loss limits per strategy are stated in dollars per 0.01 lot on the product page, and a detailed setup guide is public
  • Five public signals on real-money accounts per the MQL5 rules, at four different brokers
  • Free demo for the strategy tester

Against

  • The built-in loss limit of up to $75 per 0.01 lot already exceeds the vendor's own risk recommendation on the recommended account; if all strategies hit their stops at once, that costs over a third of the account
  • Very short track: the product has been on the market since 11 Nov 2026, the signals have run for only a few weeks and all started before the launch
  • Most of the growth on all five signals came before the market launch; on three accounts more than half even came before the official signal start
  • Very high leverage on all signals; four accounts run with a fraction of the recommended capital and the vendor himself calls them "High Exposure"
  • Almost all trades so far closed in profit while the account was at times well underwater: the track does not yet show what triggered stops do
  • With EU leverage of at most 1:20 on gold, each position ties up a multiple of the margin the signal accounts need; with several strategies open at once, small accounts get tight
  • The price is an offer price, per the vendor half the regular price, with increases planned

Buying and running

Our interests: We earn no commission on Allylla Pullback Bot and link neither the vendor nor MQL5. ATS pays us a commission when you buy through the link to the alternatives. We develop our own EA, Dawnbreak. We have no stake in Allylla Pullback Bot. The grade is set beforehand and independently of that.

Terms

Ad
Price
$299
Sold
MQL5 market only
Alternatives at ATS

We do not link to Allylla Pullback Bot itself. The affiliate link changes neither grade nor text.

What we do not know

  • How the EA copes with a losing streak.Almost all trades in the track so far closed in profit. How often the stops trigger and what several stops in a row do at 3 to 5% risk per trade is not on record.
  • Whether the stop-loss sits with the broker.Per the setup guide you can choose whether the stop sits on the broker's server or is only watched locally in the terminal. Which is the default and which runs on the signals is stated nowhere.
  • How trades are managed while underwater.The accounts were at times much further underwater than their closed history shows. The setup guide names time-based exits at breakeven or a small loss. How long losing positions stay open is not described.
  • How much margin the EA needs at EU leverage.All signals run with very high leverage. How often several strategies are open at once, and whether the margin at 1:20 on gold suffices on small accounts, they do not show.

Frequently asked questions

What is the Allylla Pullback Bot?

An Expert Advisor for MetaTrader 5 by Uttam Kumar Nandeibam that trades gold (XAUUSD) only. Per the vendor it looks for structural breaks across several timeframes and enters on retracements, with five strategies, one entry each and a stop-loss, without grid or martingale. It has been on the MQL5 market since 11 Nov 2026 and cost $299 as an offer price on 17 Nov 2026, or $149 a month to rent.

What risk does the vendor recommend?

The product page says: "The suggested risk per trade is around 3% to 5% of equity." (read 17 Nov 2026). That is high. At this risk per trade, a run of a few losses costs a large part of the account, and when several of the five strategies are open at once, the risk adds up. Anyone testing it should set it much lower. How to derive the lot size from the risk is explained in the guide on position sizing and risk per trade.

Do the signals run on real money?

Yes, as far as MQL5 ensures it: the signal rules only allow real-money accounts, and all five signal pages mark the account as a Real Account on the broker's live server (read 17 Nov 2026). Per the vendor they run the default settings. Four of the accounts, however, are much smaller than the recommended capital and so carry more risk. We have no evidence of our own.

Does the Allylla Pullback Bot use grid or martingale?

Per the product page no: the vendor describes a system with one entry per strategy, a stop-loss and trade management, without adding to losing trades. We have not seen the source code. Up to five strategies can hold positions at the same time, though. Other gold EAs by the same vendor use grid management per their product pages.

Why is the grade provisional?

Because the product has only been sold since 11 Nov 2026 and the signals have only run since 23 Oct to 7 Nov 2026 (as of 17 Nov 2026). In that time the EA has closed hardly any losing trades; how it copes with a losing streak is not on record. We will re-check the grade once the signals have run for several months or a larger decline occurs.

Is there a verified live track for Allylla Pullback Bot?

Yes, 5 public signals on real-money accounts (per MQL5 rules), selected by the vendor. Which settings run there is not disclosed. Because the vendor chooses what to show, such accounts earn at most 8.0 out of 10 for evidence. We deliberately do not show returns or drawdowns.

Changes to this review

  1. First published with a provisional grade. Product page, version history, comments, setup guide, five MQL5 signals and the MQL5 signal rules read on 17 Nov 2026.
  2. Second review by an AI agent (different model, blind): 6 of 6 criteria agree.

More commodity EAs we reviewed

Compare gold EAs

Compare MQL5 EAs