PROJECT CHIMBA R&D
Your Private Computational R&D Lab for Your Trading Strategy
Activation: $6,000 one time
Ongoing Research Capacity: $1,000/month per dedicated Research Worker
2-worker minimum after Launch Month
YOUR STRATEGY IS SOFTWARE NOW. LET'S SEE WHAT ELSE IT CAN BECOME.
You already did the hard part.
You built a trading strategy.
You defined it.
You digitized it.
You validated it.
And now your production robot can execute it without requiring you to sit in front of a chart and manually make every decision yourself.
Project Chimba R&D answers the next question:
What else could the computer discover?
Instead of asking you to manually backtest one idea at a time, Chimba R&D builds a private computational research operation around your strategy.
Dedicated Research Workers continuously generate experimental strategy configurations, simulate them against historical market data, score the results according to objectives you care about, record the experiments, identify interesting candidates and continue searching.
Your production robot executes.
Your R&D Lab experiments.
They are deliberately separate systems.
TWO SYSTEMS. TWO JOBS.
PRODUCTION
Your production robot executes the strategy you have reviewed and accepted.
It does not randomly change itself because a historical simulation looked interesting.
R&D
Your R&D Lab experiments on copies of your strategy.
It can generate alternatives.
Change parameters.
Change instruments.
Change timeframes.
Change entries.
Change exits.
Change order structures.
Change risk configurations.
Test combinations.
Simulate them.
Score them.
Remember them.
And try again.
Nothing discovered in R&D automatically touches your production account.
Research discovers candidates.
You decide whether a candidate deserves consideration for production.
YOU TELL THE LAB WHAT "BETTER" MEANS
There is no universal definition of the "best" trading strategy.
Maybe you care most about historical drawdown.
Maybe you care about ending balance.
Maybe you care about reducing transaction frequency and transaction costs.
Maybe you care about consistency.
Maybe you care about risk-adjusted performance.
Maybe you want to maximize one metric while placing hard constraints on another.
Before Chimba R&D begins searching, we establish your research objectives.
For example:
Minimize historical maximum drawdown while maintaining at least X qualified transactions.
Or:
Maximize historical ending balance subject to a maximum drawdown constraint.
Or:
Improve transaction efficiency while maintaining defined performance characteristics.
Your objective becomes part of the mathematical scoring system used by the research machinery.
You tell us what matters.
We give the machines their job.
YOU TELL THE LAB WHAT IT MAY CHANGE
We also define the permitted research space.
Your existing production strategy becomes the baseline.
From there, you determine which parts of the strategy are sacred and which parts Chimba R&D is allowed to investigate.
Depending on the strategy and approved research space, experimental changes may include:
Indicators and Parameters
EMA periods.
RSI thresholds.
Lookback periods.
Indicator thresholds.
Confirmation conditions.
Indicator combinations.
And other configurable strategy parameters.
Markets and Timeframes
Eligible instruments.
Candle granularities.
Trading windows.
Sessions.
Long-only, short-only or eligible directional structures.
Entry Architecture
Market entries.
Limit entries.
Stop entries.
Entry offsets.
Confirmation logic.
Timing rules.
Alternative combinations of entry conditions.
Exit Architecture
Fixed profit targets.
Hard stops.
Trailing stops.
Signal-based exits.
Time-based exits.
Alternative stop/target relationships.
Other supported exit structures.
Execution Architecture
Order types.
Placement rules.
Offsets.
Cancel/replace behavior.
Other supported execution methods.
Risk Architecture
Position sizing.
Exposure constraints.
Risk thresholds.
Trade-frequency constraints.
Other approved risk variables.
If you don't want something changed, we don't change it.
If you want Chimba R&D to investigate it, we can make it part of the research space.
THIS IS NOT JUST PARAMETER TUNING
Changing EMA 50 to EMA 54 is one experiment.
But the search can go much further.
What if the timeframe changes?
What if the entry changes?
What if a market order becomes a limit order?
What if a fixed take-profit becomes a trailing stop?
What if the risk structure changes?
What if the same strategy architecture is tested against another eligible instrument?
What if several of those variables change simultaneously?
As the number of mutable dimensions increases, the possible strategy space can become enormous.
Chimba R&D is designed to search computationally through spaces that would be impractical for a trader to investigate manually one experiment at a time.
YOUR PRIVATE CHIMBA R&D CLUSTER
Your research environment consists of three major components.
THE R&D MANAGER
The Manager coordinates the research operation.
It distributes research jobs.
Tracks the experiment queue.
Coordinates available workers.
Collects results.
Maintains research state.
Updates rankings.
And centralizes the operation into your private research environment.
THE RESEARCH WORKERS
Research Workers provide the computational horsepower.
Their basic job is:
Generate.
Simulate.
Measure.
Record.
Repeat.
They continuously process experiments assigned through the R&D system while your research subscription is active.
THE RESEARCH DATABASE
Completed experiments feed into a centralized private research history.
The goal is not to run millions of experiments and immediately forget them.
The work accumulates.
SEARCH YOUR RESEARCH
Where appropriate to the experiment type and storage architecture, Chimba R&D retains structured information necessary to understand and interrogate completed research.
Experiment records may contain information such as:
Experiment ID
Strategy architecture
Parameters
Instrument
Granularity
Historical period
Entry methodology
Exit methodology
Execution assumptions
Risk configuration
Transaction-cost assumptions
Trade count
Historical performance metrics
Drawdown metrics
Objective/fitness scores
Research method
Dataset/version information
Want to know whether Chimba ever evaluated a particular combination?
Search the research history.
Maybe it did.
Maybe it didn't.
But your R&D operation is designed to preserve the work it has already performed rather than repeatedly starting from zero.
DIFFERENT QUESTIONS PRODUCE DIFFERENT LEADERS
Chimba R&D does not pretend that one candidate is universally "best."
Instead, research results can be viewed according to different objectives.
For example:
If You Care Most About Drawdown
Here are the leading candidates discovered according to the defined historical drawdown objective.
If You Care Most About Ending Balance
Here are the leaders according to the defined historical simulation assumptions.
If You Care Most About Transaction Efficiency
Here are the leading candidates according to the defined transaction-frequency or cost criteria.
If You Care Most About Risk-Adjusted Performance
Here are the leading candidates according to the selected methodology.
If You Care Most About Robustness
Here are the candidates that performed most consistently according to the defined robustness methodology.
Your accepted production strategy remains available as the baseline for comparison.
The point is not to hand you a magic winner.
The point is to show you what the research machinery discovered and let you decide what deserves your attention.
THE RESEARCH SYSTEM DOES THE MATH. THE REPORT EXPLAINS IT.
Chimba's deterministic research machinery performs the simulations and calculates the results.
The reporting layer organizes those results into something a human can actually use.
Your research reporting can explain:
What was tested.
How much was tested.
What assumptions were used.
How the production baseline performed historically under those assumptions.
Which candidates led particular objectives.
What tradeoffs appeared.
Which candidates repeatedly appeared near the top.
Which apparently strong candidates failed additional testing.
What may deserve further investigation.
The research machinery produces the quantitative results.
The reporting layer communicates them.
The machines measure.
The report explains.
You decide.
THE $6,000 CHIMBA R&D ACTIVATION
A Chimba R&D subscription does not begin by simply switching on a generic optimizer.
We first build and configure your private research operation.
Your one-time $6,000 Chimba R&D Activation includes:
Private R&D Lab Buildout
Your dedicated Chimba R&D environment is provisioned and configured.
Production Strategy Integration
Your accepted digitized strategy becomes the baseline for the research environment.
Search-Space Engineering
We define what the research machinery may change.
Objective / Fitness Engineering
We translate what you care about into the research objectives and constraints used to evaluate candidates.
Historical Data Configuration
The appropriate available historical datasets and assumptions are configured for the approved research methodology.
R&D Manager
Your research control plane is configured to coordinate the operation.
Private Research Database
Your experiment history has a dedicated place to accumulate.
R&D API and Research Interface
Your research operation receives its associated access and interface as provided by the current Chimba R&D platform.
And then we turn the horsepower up.
BONUS: THE 10X LAUNCH MONTH
We don't launch your new R&D operation at the two-worker minimum.
We launch it with:
10 DEDICATED RESEARCH WORKERS.
For your entire first billing month.
We absorb the additional research capacity.
Your first month is when we're establishing the research operation, populating your experiment history and beginning to explore your approved strategy space.
We want the lab starting with momentum.
So your $6,000 Activation includes your entire:
10X LAUNCH MONTH
1 R&D Manager
10 dedicated Research Workers
Your private research database
Your configured research objectives
Your approved research space
Your first month of continuous computational research
At the end of Launch Month, you decide how much research capacity you want to keep.
AFTER LAUNCH MONTH: CHOOSE YOUR HORSEPOWER
Every dedicated Research Worker is:
$1,000/MONTH
The minimum active Chimba R&D cluster is two workers.
2 Workers — $2,000/month
3 Workers — $3,000/month
5 Workers — $5,000/month
10 Workers — $10,000/month
Or another supported worker count based on your research requirements.
There is no artificial feature wall between plans.
More workers buy more dedicated parallel research capacity.
Actual experiment throughput depends on factors such as strategy complexity, data granularity, historical periods, instruments, research methodology and experiment complexity.
Additional workers do not guarantee proportionally faster discovery or better future trading performance.
They provide something much simpler:
More computational horsepower pointed at your research problem.
RESEARCH IS SUPPOSED TO REACH PRODUCTION
There would be something fundamentally backwards about charging you to continuously discover interesting strategy candidates and then penalizing you every time the research actually produces something you want to use.
So we don't.
Every active Chimba R&D subscription includes:
TWO PRODUCTION GRADUATIONS PER MONTH.
If R&D discovers a candidate you believe deserves production after you've reviewed it and performed your own due diligence, Chimba can take that candidate through the defined Production Graduation process.
That means moving an eligible R&D-native candidate through final validation and acceptance and into the appropriate next production version.
If you find something interesting two weeks into the month and you're ready:
Graduate it.
If you find another candidate later and decide that one deserves production too:
Graduate it.
Those Production Graduations are already included in your active R&D subscription.
We do not penalize you because the research worked.
YOU DON'T HAVE TO USE YOUR GRADUATIONS IMMEDIATELY
Research should not create artificial pressure to modify a live trading system.
If nothing deserves production this month, don't deploy anything.
If you use one graduation, use one.
If you use both, use both.
24 PRODUCTION GRADUATIONS.
The goal is not to make you constantly modify your strategy.
The goal is to remove an additional Chimba deployment fee as the reason you refuse to deploy a research candidate you have independently decided deserves production.
WHAT COUNTS AS A PRODUCTION GRADUATION?
A Production Graduation applies to an eligible candidate generated within the approved Chimba R&D research space and supported production architecture.
The normal path is:
Research Candidate
↓
Candidate Freeze
↓
Final Validation
↓
Human Review
↓
Acceptance
↓
Production Graduation
↓
Production vNext
Research configurations explicitly supported by the customer's Chimba environment can therefore move toward production without creating a new deployment bill every time R&D succeeds.
A materially new software project or functionality outside the approved R&D and production architecture is not a Production Graduation and may require separate custom development.
Physical seperation
Your Chimba Production System and Chimba R&D Lab are physically and architecturally separate systems.
They run on separate servers. They run separate codebases. The R&D workers do not have access to your production environment, and the research system is not connected to the production system in a way that allows an experiment to modify or deploy itself into production.
Chimba R&D can test, simulate, rank, and report.
It cannot promote itself to production.
When you decide a research candidate deserves to become your next production version, that happens through a separate, deliberate Production Graduation process.
PRODUCTION NEVER CHANGES ITSELF
This boundary does not move.
Suppose R&D discovers a candidate with:
EMA 54
instead of:
EMA 50
and a:
limit-order entry
instead of:
market entry.
The candidate can have extraordinary historical results.
Nothing happens automatically.
It remains a research candidate.
You inspect it.
You perform your due diligence.
You decide whether it deserves further validation.
And only after deliberate acceptance can it graduate toward production.
Research proposes.
You decide.
Production executes.
THE NEW PRODUCTION VERSION BECOMES THE NEXT BASELINE
Suppose you approve Candidate #827491.
It passes the defined graduation process.
Production v1 becomes:
Production v2.
Now your R&D operation can continue.
The new accepted production strategy can become the next baseline for additional research where appropriate.
That creates a continuous R&D cycle:
Production
↓
Research
↓
Discovery
↓
Due Diligence
↓
Graduation
↓
New Production Version
↓
More Research
Your production system can remain stable for as long as you want.
Your R&D operation can keep asking questions in the background.
RESEARCH AT COMPUTATIONAL SCALE
Chimba R&D was built around the idea that computers can evaluate strategy variations at a scale that would be unreasonable to investigate manually.
The exact number of experiments varies with the research problem.
A simple parameter change is not computationally equivalent to a complicated strategy architecture evaluated across multiple markets, granularities and historical periods.
So we do not promise that every strategy will produce the same number of simulations per hour.
Instead, Chimba gives you dedicated computational workers and makes the actual work visible.
Where supported by the current interface, your R&D environment can expose information such as:
Active workers
Experiments completed
Current research objective
Accepted production baseline
Current research queue
Leading candidates
Research history
Worker activity
Searchable experiments
Available Production Graduation credits
Research reports
You don't have to take our word for how much work the machinery is doing.
The goal is to let you see it.
WE SEARCH. WE DON'T PROMISE WHAT WE'LL FIND.
Historical optimization has a dangerous property:
If you torture historical data long enough, you can create extremely attractive historical results that mean very little.
Chimba R&D is not sold as a machine that guarantees a better trading strategy.
Research methodology may therefore include, where applicable:
Search / Training
Generate and identify candidates.
Out-of-Sample Validation
Evaluate interesting candidates against data not used to discover them.
Robustness Testing
Challenge candidates across defined periods, markets, assumptions or other relevant conditions.
Ranking
Compare surviving candidates according to your selected objectives.
The goal is systematic computational research.
Not the prettiest backtest we can manufacture.
WHAT CHIMBA R&D DOES NOT GUARANTEE
Chimba R&D does not guarantee:
That a better strategy will be discovered.
That any research candidate will outperform your existing strategy in future markets.
That historical performance will repeat.
That your strategy will become or remain profitable.
That a particular return will be achieved.
That future drawdown will decrease.
That adding additional Research Workers will create proportionally better research discoveries.
Sometimes the research may tell you:
We investigated this strategy space extensively and did not find a candidate sufficiently compelling to justify replacing the current baseline.
That is still a research result.
THE CHIMBA R&D COMPUTE GUARANTEE
We guarantee what we actually control.
While your paid Chimba R&D subscription is active, Chimba provides the dedicated Research Worker capacity included in your active subscription and operates the agreed research process against your configured research environment, subject to the service terms, scheduled maintenance, third-party infrastructure availability and agreed technical scope.
If a Chimba-controlled implementation failure materially prevents purchased research capacity from operating as agreed, Chimba will correct the service issue and provide the applicable remedy defined in the service agreement.
We guarantee the research machinery.
We do not guarantee what the markets will reveal.
MONTH-TO-MONTH AFTER ACTIVATION
Your $6,000 Activation includes your first month and the 10X Launch Month.
After that, you choose how much of the lab you want to keep running.
The minimum is:
2 Research Workers — $2,000/month.
Want more?
Add workers at $1,000/month each.
Want fewer than the ten you experienced during Launch Month?
Scale down.
Want all ten to keep working?
Keep all ten.
Want even more computational capacity?
Talk to Chimba about expanding the cluster.
As long as your R&D subscription remains active, your selected research capacity continues working according to the service terms.
Your production Strategy-to-System is separate.
Ending or reducing your R&D subscription does not mean Chimba gets to take away the production system you already own.
WHO CHIMBA R&D IS FOR
Chimba R&D is for someone who already has a digitized systematic trading strategy and understands what computational research can and cannot prove.
It's for the trader who looks at their production strategy and asks:
"What if?"
What if we changed the exit?
What if we changed the entry?
What if we changed the timeframe?
What if we changed the instrument?
What if we changed the risk model?
What if we changed all three?
What if we optimized around drawdown instead of ending balance?
What if we let computers investigate this every day instead of me trying another backtest whenever I have time?
If those questions don't interest you, you probably don't need Chimba R&D.
If they do:
That's why we built the lab.
CHIMBA R&D
$6,000 ACTIVATION
Includes:
Private R&D Lab buildout
Strategy research integration
Search-space engineering
Objective / fitness engineering
Historical-data configuration
Private R&D Manager
Research database
R&D API and interface
10X Launch Month
10 dedicated Research Workers for your first month
2 Production Graduation credits during Launch Month
Then:
$1,000/MONTH PER RESEARCH WORKER
2-worker minimum
Every active month includes:
2 ADDITIONAL PRODUCTION GRADUATIONS
Unused eligible graduations may accumulate up to 24.
YOUR PRODUCTION ROBOT HAS ONE JOB:
EXECUTE WHAT YOU APPROVED.
YOUR R&D LAB HAS ANOTHER:
NEVER STOP ASKING "WHAT IF?"
You decide what it can change.
You decide what better means.
Chimba supplies the machinery.
And when the lab discovers something you believe deserves production:
We want you to use it.
That's why Production Graduations are built into the subscription.