• Welcome back! Thank you for being a part of this Traders Community. Let's discuss and share :)
    Selamat datang kembali! Trimakasih telah menjadi bagian dari Komunitas Trader ini. Mari berdiskusi dan berbagi :)

New EA DeepGold LLM EA – when gold listens to artificial intelligence (well, almost)

DeepSeek, like a real star, decided to rebrand itself — and our good old deepseek-chat model went on a well-deserved vacation, and it was replaced by deepseek-v4-pro and deepseek-v4-flash.

The adviser, not understanding the humor, tried to reach the old model and received an HTTP 400 response like "guy, where are you calling, there is no such model anymore."

What has changed:

Changed the model name in the code from deepseek-chat to deepseek-v4-pro.

Now the AI is back in touch, smarter than before and ready to predict gold movements, rather than drawing errors in logs.

If you have an older version, just download the updated version. Now the advisor is back online, and DeepSeek has stopped pretending to be unavailable. Profit for everyone and no 400 mistakes!
 

Attachments

NPNavigator said:
Hello, colleagues!

Now our "golden" AI advisor has stopped being stubborn as a sheep and has learned to obey his master.

What was added:

• Two new fields in the settings: LLM_BuyThreshold and LLM_SellThreshold

(search in the "AI Settings (DeepSeek)" group).

Previously, the thresholds were tightly sewn in:

prediction >= 0.65 → BUY, prediction <= 0.35 → SELL.

Now you decide for yourself with what confidence the AI should open a deal.

If you want, tighten the nuts to 0.8 and 0.2 in order to trade only "reinforced concrete" signals.

If you want, loosen to 0.6 and 0.4 and catch more movements (but with popcorn, because it can be stormy).

Now, instead of "the robot thinks you're a sucker and ignores your wishes," we have a "henpecked robot" - turn the sensitivity knob and watch as it obediently runs to open positions.

Try it, experiment, and share the results. Version 1.4 is already in the topic header. Profit for everyone!
Click to expand...
hello
i am facing "HTTP 400" error, removed EA from the chart and added again, but still same issue. please give me solution
 
For gold robots, I usually separate the signal logic from the execution assumptions. The same idea can look solid in a backtest, but I would want to compare it across spread regimes, then check out-of-sample behavior and forward-test drift. Session timing matters too, especially when liquidity thins out or the broker widens spread around news. If the result changes a lot when those inputs move, the edge is probably too fragile.
 
Otabek said:
hello
i am facing "HTTP 400" error, removed EA from the chart and added again, but still same issue. please give me solution
Click to expand...
The HTTP 400 error is not a glitch, it's DeepSeek that has granted an "amnesty" to old models and retired deepseek—chat. The counselor knocks on her door out of habit, and she says, "There's no such model anymore, come back tomorrow."

The full detailed analysis of this problem lies in message No. 62 — there I tell you what to replace the model with so that everything starts up again.

So scroll a little higher, read post No. 62, download the updated version — and let the gold work for us again, not on the log screen!
 
samshoshoEA said:
For gold robots, I usually separate the signal logic from the execution assumptions. The same idea can look solid in a backtest, but I would want to compare it across spread regimes, then check out-of-sample behavior and forward-test drift. Session timing matters too, especially when liquidity thins out or the broker widens spread around news. If the result changes a lot when those inputs move, the edge is probably too fragile.
Click to expand...
I am also in favor of having the logic of signals and execution live in different "apartments" — this makes it easier to understand who is messing up: the predictor neural network or the crooked hands of a broker with an extended spread on the news.

Everything looks beautiful on the backtest — the graph goes into space, no martingales are needed. But once you turn on a realistic spread, add slippage, and remove the Asian session with its ghost spikes— the Grail turns into a pumpkin.

The Expert Advisor already has a session filter and a spread check — this is the first echelon of defense against market brutality.

Comparing the behavior on different configurations is like a test drive of a car in ice and rain: if it drifts on a wet road, it means that the suspension is not very good, and the advantage is really fragile.

In short, the idea is sound — dig deeper, test harder.
 
NPNavigator said:
The HTTP 400 error is not a glitch, it's DeepSeek that has granted an "amnesty" to old models and retired deepseek—chat. The counselor knocks on her door out of habit, and she says, "There's no such model anymore, come back tomorrow."

The full detailed analysis of this problem lies in message No. 62 — there I tell you what to replace the model with so that everything starts up again.

So scroll a little higher, read post No. 62, download the updated version — and let the gold work for us again, not on the log screen!
Click to expand...
in new version of 1.5 i faced a new error: HTTP 1003, what doest that mean ?
 
Otabek said:
in new version of 1.5 i faced a new error: HTTP 1003, what doest that mean ?
Click to expand...
Error 1003 from DeepSeek means: "You're being too persistent, chill out." Literally, your account or IP address is temporarily blocked due to exceeding the request limits or violating the API terms of use. You may have accidentally staged a DDoS attack with your experiments.

What to do:

1. Check the balance and quotas in DeepSeek's personal account. It's possible that you just ran out of free tokens.

2. If you have several Expert Advisors or copies on different charts, they all break into the API at the same time, and DeepSeek turns on the bastion mode.

3. Wait a bit (usually the unit is removed after 15-30 minutes) and try again.
 
NPNavigator said:
Error 1003 from DeepSeek means: "You're being too persistent, chill out." Literally, your account or IP address is temporarily blocked due to exceeding the request limits or violating the API terms of use. You may have accidentally staged a DDoS attack with your experiments.

What to do:

1. Check the balance and quotas in DeepSeek's personal account. It's possible that you just ran out of free tokens.

2. If you have several Expert Advisors or copies on different charts, they all break into the API at the same time, and DeepSeek turns on the bastion mode.

3. Wait a bit (usually the unit is removed after 15-30 minutes) and try again.
Click to expand...
thanks, i was using paid tokens, but i each time i created a new api key and now for 1.5 version i was using the third api key since we started testing DeepGold EA
 
Otabek said:
thanks, i was using paid tokens, but i each time i created a new api key and now for 1.5 version i was using the third api key since we started testing DeepGold EA
Click to expand...
The third API key in a short time? Yes, you are already a VIP customer of DeepSeek! With such an expense of keys, you will soon be given branded mugs and calendars.

But seriously, you may not be burning your keys, but simply losing them due to locks. When the server receives a flurry of requests from one account (even with a new key), it still sees who is knocking, and it can no longer ban the key, but the entire account or IP.

Tip: don't change your keys like gloves, but look into DeepSeek's personal account and check the request limits per minute (RPM) and per day (TPD). Perhaps you are just hitting the ceiling of the paid tariff, and you need to either expand the quota, or pull the neural network a little less often.

By the way, the Expert Advisor has SpikeCooldownSec = 300, which is the pause between requests for spikes. If you reduced it to 5 seconds for the sake of "super-fast signals", the AI server simply turned on the "do not disturb" mode

So: one key, proven limits, a little patience, and DeepSeek will be your best friend again, not a moody star!
 
Hi,
same problem
I change the ea to the new one and today i have error HHTP 1003 on API Statut.
the key used is the same than used with previous version and solde on Deepseek is always positive.
 
NPNavigator said:
DeepSeek, like a real star, decided to rebrand itself — and our good old deepseek-chat model went on a well-deserved vacation, and it was replaced by deepseek-v4-pro and deepseek-v4-flash.

The adviser, not understanding the humor, tried to reach the old model and received an HTTP 400 response like "guy, where are you calling, there is no such model anymore."

What has changed:

Changed the model name in the code from deepseek-chat to deepseek-v4-pro.

Now the AI is back in touch, smarter than before and ready to predict gold movements, rather than drawing errors in logs.

If you have an older version, just download the updated version. Now the advisor is back online, and DeepSeek has stopped pretending to be unavailable. Profit for everyone and no 400 mistakes!
Click to expand...
Hi
I want to ask , is this can run on cent account just for the real test ?
 
It seems that we have not an isolated case here, but a direct flash mob "HTTP 1003"! The DeepSeek server apparently decided that Doomsday had come and turned on the "fortress siege" mode for everyone at once.

It happens when DeepSeek has a global glitch, or they massively shake up access rules, and half of the users become "suspicious" simply for breathing.

What to do while they fix their neurons there:

• Don't change keys like crazy — it won't help, you'll only hit the limits. One key for everything.

• Check the SpikeCooldownSec in the settings — if it is too small (like 5 seconds), the server thinks that you are attacking it. Hold at least 300 (5 minutes) — let the neural network digest the previous query.

• Try a VPN or IP change — sometimes Cloudflare bans a specific range of addresses, not you personally. Moving to another country can virtually solve the problem.

• Look at the DeepSeek status (if they have a status page) - perhaps they are honestly writing there: "We've fallen, wait."

Well, remember: if the neural network does not respond, it does not mean that it does not love you. It's just that it's "Monday" for her. We are waiting for DeepSeek to stop acting up. Patience and green candles for everyone!
 
NPNavigator said:
Hello, colleagues!

I used to post quite classic Expert Advisors here – indicator, martingale, grid… But this time I was carried away towards neural networks and large language models. Why? Yes, because there are no such advisers on the forum, and the topic is wildly interesting. I wanted to understand if LLM (in my case DeepSeek) can really help in trading, or if it's just a HYPE.

This is how DeepGold LLM EA was born, a robot that sends a "market snapshot" directly to the DeepSeek cloud neural network every 15 minutes (and with sudden movements), and it returns a forecast: buy, sell or stand aside. And yes, he trades only in gold (XAUUSD) and strictly on M15 – no amateur activity.

What's under the hood? Key features

• A living AI brain - the Expert Advisor collects a lot of data: ATR, ADX, RSI, CCI, Bollinger bands, volumes, the last 30 candlesticks of M15, open positions, the current P&L, and even the history of its past forecasts with a "guessed/not guessed" rating. All this goes to the DeepSeek server, and the model returns a number from 0 (furious SELL) to 1 (confident BUY) along with a confidence score.

• Protection against "nonsense" – if the confidence is below the set threshold (65% by default), the transaction is not opened. This is how we cut off weak signals.

• Dual stop system – you can use real StopLoss / TakeProfit or virtual stops (so as not to give them to the broker), as well as a trailing stop, taking into account the average price of all positions.

• Smart risk management – the risk per trade is calculated from the balance, there is a daily loss limit (by default, 5% – and trading freezes until the next day), maximum spread, Spike Filter, protection against cascading entry through the ATR distance.

• Schedule and timezones – you can set the UTC trading hours (so as not to get into the Asian "mess"), as well as compensate for the server time offset.

• Own dashboard – a window with the current signal, number of orders, AI confidence, API status, ATR, spread, equity and drawdown is displayed on the chart. Everything is clear, without unnecessary graphics.

• Self–learning (almost) - the EA remembers its forecasts and checks after a set number of bars whether they have come true, transmitting this feedback in the following queries. This is how the model sees its mistakes and adjusts.

• Live only – it won't work in the strategy tester (I specifically blocked it), because it's pointless without a real connection to the API.

Recommendations for use (highly recommended!)

1. Start with a demo account. Even if you are an MQL5 guru, this advisor is an experiment. Let him run for a couple of weeks without risking any hard money.

2. Get the DeepSeek API key. They give you a decent amount of free tokens, but if you want to actively send requests, you may need to top up your balance. Insert the key into the parameters.

3. Adjust the trading time to your UTC. The default time is 9-23 UTC, which roughly covers the European and American sessions. Check if the server time matches.

4. Take a risk on the deal – don't be greedy. 1% at a stop of 400 pp is quite a working option for the test. If you set 5% or higher, the daily limit will work quickly in case of a series of losses.

5. Do not change the timeframe and symbol. The robot is rigidly attached to the XAUUSD M15, it will simply refuse to work on other instruments (it will issue an Alert).

6. Keep an eye on the instrument panel. If the API status is red, check the Internet or the token balance.

7. The Advisor is not the "Grail". This is an assistant that uses a probabilistic approach. The model can be wrong, the market is an unpredictable thing. Treat it like an advanced assistant, not like a magic "loot" button.

The advisor is accompanied by detailed instructions describing each parameter and the logic of its operation. Download, try, and share your impressions. I will be glad for any feedback, especially from those who are already experimenting with LLM in trading.

If you have any ideas on how to improve the "brains" – write. Perhaps we will add support for other providers (OpenAI, Claude) or more sophisticated strategies.

Good luck and profits!
Click to expand...
 
The situation is this: deepseek-v4-pro sometimes turns on the "stars" mode — it gets cranky and throws HTTP 1003, saying, "don't come, I'm tired." And her younger sister, deepseek-v4-flash, plows like a caffeine—fueled workaholic - flawlessly and smartly.

So now I'm preparing a new version of the Expert Advisor, where you can choose a model right in the settings: if you want a smart but thoughtful Pro— you're welcome. If you want fast and stable Flash— no problem. One click and no problems with the code.

As soon as I finish testing and I'm sure that nothing falls off, I'll post an updated Expert Advisor.
 
You must log in or register to reply here.
Back
Top