MonarqiLabs is a stress-testing simulator for the Indonesia Stock Exchange. It runs a controlled market of 961 real IDX tickers and ten AI personas with independent memories. You inject one shock and watch what the market would do, day by day, against a counterfactual baseline.
A definition in plain words, and the line between simulation and prediction.
MonarqiLabs answers one question: what would the Indonesian stock market actually do if a specific shock hit it? Not what should it do, not what did it do last time, but what a market built from 961 real IDX stocks and ten different trading behaviors would do next.
Every crash looks obvious in hindsight. The desk's premise is that hindsight is the wrong teacher: it only shows you the path that happened. A simulation shows you the mechanism: who sells first, who buys the dip, how fast panic spreads, and where the price finally settles. Watching that happen is the product.
Two limits keep it honest. First, prices after a shock are scenarios, not forecasts: the starting valuations are grounded in the real IDX, but what follows is simulated behavior. Second, it is a research and teaching tool, not investment advice. No real money moves anywhere on this site.
One shock in, 120 simulated trading days out, with a counterfactual baseline behind every price.
A run starts from a scenario. Each scenario names its shock, when it lands, how severe it is, and which stocks it touches. From there the market runs itself for the scenario's length (most are 120 trading days), and the desk records every event to a log.
On the scenario's trigger day the shock hits: a rate cut, an index flush, a policy surprise. Its severity scales the price pressure it applies.
Each of the ten personas reads the shock through its own strategy and writes a reasoning line. The same news produces ten different opinions.
Opinions become orders. Every order names its agent, stock, side, quantity, and reason, and every fill is recorded with the price paid.
Fills move prices, prices change what looks cheap or stretched, and the loop repeats the next day. That feedback is what makes a run a market instead of a chart.
The same market, same seed, no shock. That is the counterfactual baseline. The desk replays both, so the view always separates what the shock changed from what the market did anyway.
Every scenario ships with a fixed seed (42 in the curated library). Same scenario, same seed, same log, byte for byte. A run you open today will match the run you open next month, which is what makes logs linkable and discussion possible.
Ten independent trading behaviors, each with its own memory and its own opinion of every event.
The personas are the market. Each one follows a single strategy, holds a rolling memory of what it has seen, and reacts to news and shocks in character. A rate cut is good news to Macro Marge and arithmetic to Value Vera; a red day is a discount to Contrarian Carl and a signal to exit for Panic Pete.
| Persona | Style | What it does |
|---|---|---|
| Value Vera | Value | Buys what trades below intrinsic value. Sells what is rich. |
| Momentum Max | Momentum | Buys 20-day moving-average breakouts. Exits before breakdowns finish. |
| MeanRev Mia | Mean-Reversion | Fades 20-day z-score extremes: buys stretched-down names, trims stretched-up ones. |
| FOMO Fred | FOMO / Herd | The herder: chases two-day winners, trims anything up 5%. |
| Panic Pete | Panic / Defense | Sells into his own drawdowns. Cash is comfort. Cuts hard, cuts early. |
| News Nelly | News Desk | Trades the tape: buys relative-strength leaders when sentiment is positive, de-risks when it is negative. |
| Contrarian Carl | Contrarian | Fades big up-days, catches big down-days. Crowds are the counterparty. |
| Passive Pat | Control / B&H | Buy-and-hold equal-weight basket from Day 1. Never re-trades. |
| Fund FM | Risk Mgr | Strict stop-losses and volatility-scaled position sizes. Process over prediction. |
| Macro Marge | Macro | Trades the rate proxy: longs when rates fall, defends when they rise. |
23 scenarios, each anchored to something that actually happened on the IDX.
Every scenario in the library is anchored to a real event or a recurring real pattern on the Indonesia Stock Exchange. That anchor is written into each scenario's description, so a run is always a comparison: this is what happened, and this is what the simulated market did with the same shock.
| Scenario | Shock | Real anchor |
|---|---|---|
| MSCI Free-Float Freeze | Passive funds de-risk ahead of a possible exclusion wave | Oct 2025: MSCI's new free-float methodology hit the IHSG from its 8,355 high; the Jan 27, 2026 freeze drove an 8% intraday drop with a trading halt and ~$2.2B expected outflow. |
| MSCI Quarterly: Indonesian Additions | ETFs must buy at the rebalance date | Nov 2025 review: BREN and BRMS added to MSCI Global Standard; inclusion events averaged ~$247M inflows per name, BREN +244% around the window. |
| MSCI Quarterly: Deletion Wave | Forced ETF selling | The May 2026 review removed six Indonesian companies including Amman Mineral and Barito Renewables; IHSG -3.5% over the week. |
| BI Rate Cut 25bps | Cheaper funding lifts bank margins | Aug 19, 2025: BI-Rate cut to 5.00%, the fourth cut of 2025. Event studies on 41 banks found no significant abnormal return: the cut was fully anticipated. |
| BI Rate Hike (Rupiah Defense) | Costlier funding dents bank NIMs | 2024: BI held or hiked amid rupiah weakness toward 16,700/USD; a higher BI-Rate is a negative signal for bank stock returns in the panel data. |
| Finance Minister Reshuffle | Markets test fiscal credibility overnight | Sep 8-9, 2025: the finance minister was replaced; IHSG fell ~1.6% on the day after an intraday record 8,022. |
| Budget Efficiency Shock | Austerity or reallocation, unclear | 2025 'budget efficiency' program: state revenue shortfalls plus the ambiguity drove IHSG weakness in H1 2025. |
| SOE Export Mandate | Policy applied to state exporters | May 21, 2025: IHSG -3.54% after the SOE export policy at the DPR; energy led the drop at -6%. |
| Presidential Speech Blunder | A misread statement triggers a dip | Modeled on recurring 2024-25 headlines: a 1-2% intraday dip that mostly reverses within the week. |
| Coretax Revenue Miss | Revenue below target on a rushed system | Mid-March 2025: IHSG sold off after Jan-Feb state revenue came in below target, driven by the rushed Coretax reporting rollout. |
| Rupiah Crisis Mode | Currency slides, importers squeeze | 2024-25: the rupiah slid toward 16,700-16,800/USD on MSCI and tariff stress; exporters tailwind, importers and debt-funded property headwind. |
| US Reciprocal Tariffs | External trade shock, regional rotation | Apr 2, 2025 'liberation day': IHSG fell 8% through the 90-day pause window while regional peers did worse; IHSG ended H1 2025 -17%. |
| Oil Price Spike | Import squeeze, commodity tailwind | Commodity cycle: coal-linked earnings (ADRO/PTBA) track prices; an oil spike squeezes importers and consumer names and is a headwind for the rupiah. |
| Gold Rally Safe Haven | Safe-haven rotation into emitters | 2024-25: gold above $2,600/oz lifted gold-related emitters during tariff and MSCI stress. |
| Banking Blue-Chip Rally | Liquidity concentrates in banks | 2024: BBRI EPS growth 78% YoY; BBRI/BCA/BBNI/BMRI drive a large share of IDX volume. |
| Property Sector Recovery | Rate-sensitive but lagging | 2025: four BI cuts through 2025, yet M2 liquidity stayed thin and the big four property names (BSDE, CTRA, SMRA, PWON) underperformed. |
| Earnings Season: Broad Beats | Positive surprises lift the leaders | Quarterly pattern: positive surprise season historically lifts momentum strategies and bank-heavy blue chips. |
| Earnings Season: Guidance Cuts | Rotation to unaffected sectors | Quarterly pattern: guidance-cut season rotates money into unaffected sectors and triggers value-style buying of depressed blue chips. |
| Intraday Trading Halt | Circuit breaker, then overreaction | Mar 18, 2025: IHSG -5.02% triggered a halt; the Jan 27, 2026 MSCI freeze did the same on an 8% drop. LQ45 event studies show the overreaction mostly recovers. |
| Retail FOMO: Gorengan Run | Parabolic retail rallies, violent unwind | Recurring pattern cited in 2025: 'gorengan' stocks (low float, concentrated) build parabolic retail rallies that unwind violently when MSCI or OJK signal cleanup. |
| Election / New-Administration Rally | Growth bids infrastructure and banks | 2024 election pattern: LQ45 event studies around the election show positive abnormal returns; new-administration bids run to infrastructure and banks. |
| DPR Demonstrations | Political-uncertainty dip | Aug 29, 2025: DPR demonstrations on allowances; IHSG -2.27% the session. EM political-uncertainty dips run 5-10% when they touch policy credibility. |
| Baseline (No Event) | Counterfactual control | No shock. The market runs on persona behavior alone; this is the comparison line behind every other scenario. |
Every run ships its full event log as a downloadable JSONL file: one JSON object per line.
A log is the run's complete history. Each line is one event with a timestamp and a type,
and the desk replays from this file, so what you watch and what happened are the same
bytes. Logs live at /logs/<scenario>_seed_42.jsonl under the site root,
and the desk's run picker links straight to them.
| Event | Meaning |
|---|---|
| run_start | Run identity: seed, day count, model fingerprint. Determinism starts here. |
| day_start / day_end | Day boundaries; day_end carries the equity of all ten personas. |
| fundamentals | A stock's true value for the day, before any crowd moves it. |
| rate | The interest rate the macro side of the market reads. |
| shock | The scenario event, with type, severity, and target stock if any. |
| thought | One persona's reasoning: its read of the event, confidence, and resulting stance. Ten of these per event. |
| order | An intent to trade: agent, stock, side, quantity, price, reason, and whether it came from a rule or a shock reaction. |
| fill | An executed order at the actual price paid, with the P&L delta. |
| skip | An order the risk layer rejected, with the reason (position caps, stop rules). |
| price | The printed price for a stock after that round of trading. |
| portfolio | Each persona's cash, equity, P&L, and holdings for the day. |
Each scenario also ships a baseline_events.jsonl: the same market, same seed,
no shock. The desk's whole trick is subtracting one from the other, so the question a log
answers is never "what happened" but "what did the shock change".
The questions people actually ask, answered plainly.
No. MonarqiLabs is a market simulator. No real money moves, no orders reach any exchange, and nothing on the site is investment advice.
The 961 tickers and their starting valuations are grounded in the real Indonesia Stock Exchange. What happens after a shock is simulated by the ten personas reacting to each other, so the price paths are scenarios, not forecasts.
Yes. Every page, scenario log, and visualization is open in the browser with no account, no paywall, and no install.
There is no public API yet. The scenario logs are downloadable JSONL files, but the site does not expose a supported API today. If that changes, this page will document it first.
No. MonarqiLabs is a research and teaching tool. A simulation shows what a set of behavioral rules would do; it cannot know your situation and does not make recommendations.