NinjaCoding @ NinjaTrader
Open in Telegram
935
Subscribers
No data24 hours
No data7 days
No data30 days
Data loading in progress...
Similar Channels
Tags Cloud
No data
Any problems? Please refresh the page or contact our support manager.
Incoming and Outgoing Mentions
---
---
---
---
---
---
Attracting Subscribers
December '24Dec '24
December '24
+22
in 0 channels
November '24
+31
in 0 channels
Get PRO
October '24
+30
in 0 channels
Get PRO
September '24
+31
in 0 channels
Get PRO
August '24
+43
in 0 channels
Get PRO
July '24
+39
in 0 channels
Get PRO
June '24
+41
in 0 channels
Get PRO
May '24
+39
in 0 channels
Get PRO
April '24
+39
in 0 channels
Get PRO
March '24
+36
in 0 channels
Get PRO
February '24
+564
in 0 channels
| Date | Subscriber Growth | Mentions | Channels | |
| 21 December | 0 | |||
| 20 December | +2 | |||
| 19 December | 0 | |||
| 18 December | 0 | |||
| 17 December | +2 | |||
| 16 December | +1 | |||
| 15 December | +2 | |||
| 14 December | +5 | |||
| 13 December | 0 | |||
| 12 December | +2 | |||
| 11 December | 0 | |||
| 10 December | +1 | |||
| 09 December | +1 | |||
| 08 December | +1 | |||
| 07 December | +1 | |||
| 06 December | 0 | |||
| 05 December | +1 | |||
| 04 December | +2 | |||
| 03 December | 0 | |||
| 02 December | +1 | |||
| 01 December | 0 |
Channel Posts
| 2 |  Parallel WFO periods running all at the same time! | 119 |
| 3 |  Basically, now with MCP I am just chatting to Claude. I am working on a strategy, and I did a survey test on it last week, where I ran it on like 25 symbols with defaults. Today I came back and I asked Claude what I did, and what I decided to do next with it, and what the results were. So he reminded me that it did quite well on MES and that we decided to take it to WFO in our next session. So then we talked about the best period and parameters to be set up and I just asked him to set up a test for me which he can now either run himself or I can do it manually. Either way, this workflow now is much more persistent. | 119 |
| 4 | MCP AlgoDesk optimizer use case -> Before: create a WFO set up, create periods, plug in the strategy, set increments for all parameters you want to optimzize, plug in the correct data etc etc etc, rough time only to set the test up is 10-15min, now MCP server does this from chat in less than 30 sec | 101 |
| 5 | https://ninjacoding.net/blog/validationlifecycle | 132 |
| 6 | Looking through my old blog posts today made me feel like a dinosaur. Some of them explained code, little tips and tricks β so sweet π
Theyβll now stay there as legacy. Maybe someday my ancestors will print them out on digital holographic paper and read them out loud. | 146 |
| 7 |  Way to go with testing portfolios. After each run talking back to Claude discussing the performance and next steps makes analysis level way higher. | 316 |
| 8 | Before everyones headache was how to code a strategy, now its a matter of 1 hour. But validation is something we will all bump into. Proper and thoughtful methodology that will allow to filter out robust strategies is gold, especially if automated with MCP. | 296 |
| 9 |  No text... | 281 |
| 10 | What really becomes interesting now, is that now, with ability to integrate MCP into trading, backtesting and optimization software it is possible to really dig into things that were simply not available technically and even physically before. Standard retail trader now has access to phd level guided analysis and extensive methods of automation and research. I think the edge that AI has give traders now is enormous! | 270 |
| 11 |  +1Trying to adopt some of https://epchan.com/ Ernest Chan techniques described in Algorithmic Trading Winning Strategies and their Rationale | 250 |
| 12 | Platforms that wont integrate MCP into their workflows will die soon | 238 |
| 13 | In a matter of a few months time everyhting changed so much that now the workflow looks like this. Step 1 - lets have a look at this quant book from respectable author and see if we can make a strategy out of it that will suit my style Step 2 - ok looks good, lets make the specs Step 3 - allright code it Step 4 - put it in a backtester on a list of micros, verify it executes correctly and lets see if there is edge Step 5 - put it in WFO, lets see if its robust Step 6 - lets montecarlo some of the best takes before we go live Step 7 - ok this strategy is ready to go into our portfolio, lets stick it into live chain | 245 |
| 14 | π₯ NC Smart Scalper V1.8 is here β biggest update yet!
We've been busy. Here's what's new:
β‘οΈ NEW: Stop Breakout entry mode β don't wait for the pullback. A stop-market order sits at the forming bar's close level and fills the instant the brick completes. Switch between Pullback and Breakout live from the toolbar.
π NEW: REVERSE button β one-shot reversal hunting. Arm it, and the strategy waits for the trend flip, takes the reversal entry, then goes right back to normal scalping.
π¨ Completely redesigned toolbar β modern chip-style buttons, live scalp counter (0/2), RESET button, and Max Scalps editor right on the chart. No more digging into strategy settings mid-session.
π Draggable mark lines β grab and move your long/short zone lines while the strategy is running. Delete a line to clear it. The logic follows the line in real time.
π‘ Bulletproof order handling β new internal state machine eliminates double entries in fast markets. Entries are strictly one-at-a-time, every fill gets its profit order, every scalp counts.
π― Target Offset β fine-tune your profit target by ticks for faster fills.
π¦ V1.8 is live in your member area β update now. π₯· NC Smart Scalper: https://ninjacoding.net/strategy/smartscalper π All strategies: https://ninjacoding.net
π₯ Full walkthrough video: https://youtu.be/D-VvXwn1aiw | 460 |
| 15 |  Added a few more controls to the toolbar for SmartScaler 1.5 today, please find in member area downloads. | 377 |
| 16 | π³ You voted β I delivered!
NC Smart Scalper won the poll, so it's the first strategy to get the AI treatment. The video is out now π¬
In it, AI and I: π Review the existing strategy logic (the full setup, from zero) βοΈ Carefully modify the business logic β step by step, nothing blind π₯ Build a proper on-chart UI on top
Full process on screen, so you can repeat it with any of the 14 open source strategies.
π¦ The updated Smart Scalper code is already released in your member area β grab it and load it up.
π₯ Watch: https://youtu.be/wGVwq4tihEM?si=iqsVFKkSFp_azfof π₯· NC Smart Scalper (source included): https://ninjacoding.net/strategy/smartscalper | 378 |
| 17 | π NC-Screener is LIVE
Your whole NinjaTrader watchlist. One click. Screened in seconds.
β
Close up/down, MA, Inside Bar, Alligator, Range, 52-week highs & lows
β
Any timeframe β 1m to weekly
β
Runs natively inside NinjaTrader 8
And here's the twist: I extend it on camera using AI β and you can too. Watch me add a feature in minutes, no coding experience needed:
βΆοΈ https://youtu.be/ZAGzcP8wg5M
Get it β subscribe and use it as-is, or grab the source code and build on it yourself (setup guides + troubleshooting included):
π₯· https://ninjacoding.net/addon/ncscreener
π¬ Comment on youtube the feature you want next β I'll build it live. | 334 |
| 18 | Why my strategy research used to end in nothing β and what fixed it
For years my process looked like this: build a multi-combinatory system, throw millions of parameter combinations at the optimizer, and hunt for the ones that "work."
Sounds like research. It isn't. It's sampling noise.
Run enough combinations and pure chance guarantees some look brilliant in-sample. You can't tell the lucky ones from the real ones, so you chase ghosts β they shine in the backtest, crumble out-of-sample, and weeks evaporate. The search space grows exponentially; the information per iteration approaches zero.
The fix turned out to be one sentence:
An edge hypothesis must be falsifiable.
Before I code anything, I now write down: the condition (what setup), the expected behavior (what price should do after it), and β the key part β what test result would prove me wrong.
"Breakouts work on momentum instruments" is not a hypothesis. It can't fail, so it can't be researched β only endlessly excused.
"After X happens, price continues often enough that entering with a stop at Y is profitable before costs" β that can be killed by data. And a hypothesis that can be killed can also be confirmed.
This flips the whole workflow:
β Old: search millions of combos β find something that scores β invent an explanation after
β
New: state one specific claim β test it raw, default settings, fixed basket β it survives or it dies
Optimization still exists β but only at the end, walk-forward, tuning an edge that already proved it has signal. Search tunes an edge. Search never finds one.
Iterations went from millions to dozens. Every one ends with a written verdict β including the kills. A dead idea with a recorded reason is progress. A folder of "promising" optimizer runs is not.
Explain first. Test second. Kill fast. π§
I'm now building this whole pipeline β idea β falsifiable hypothesis β screening β walk-forward β SIM, with hard kill-gates at every step β directly into my platform. Will show you how it looks soon. π | 478 |
| 19 | So imagine running a WFO like this, then analysing the results, makeing a few optimization settings tweaks and a few strategy tweaks and then running it again and analysing two different runs against each other. In real life, for human brain its very difficult to make an objective decision what exactly became better and what became worse. Now we have AI to analyse this and talk back to you so and the EOD you get the best possible solution for you live trading. | 490 |
| 20 |  WFO survival) | 449 |
