FRERET RESEARCH
YOUR WORKSPACE / 01
Portfolio desk.
No portfolio yet
Record your first paper transaction to build your holdings.
| Stock | Shares | Avg. cost | Reference price | Value | Unrealized P/L |
|---|
USD · weighted-average cost. Enter dated reference prices to value positions. No broker orders are placed.
| Stock / action | Shares | Price | Gross amount | Executed | Saved | By |
|---|
Your allocation SPY
—Constant weights · daily rebalancing
| Asset | Weight | Annual return¹ | Volatility² | Risk share² |
|---|
¹ Historical arithmetic mean × 252. ² From the selected covariance model. USD assumed; no currency conversion.
Return relationship
Choose two assets to begin.
Run a portfolio analysis first.
Correlation uses sample returns. Covariance follows your selected estimator. Both use the same common price dates.
Efficient frontier
Model allocations
Long-only, fully invested, position cap enforced. These are in-sample estimates, not a backtest of a strategy chosen before the period.
Portfolio vs benchmark excess returns
OLS · ONE FACTOR
What moves your portfolio?
The slope measures benchmark sensitivity. Alpha is the intercept after deducting the daily risk-free return.
HC3 heteroskedasticity-robust standard errors. Alpha is annualized arithmetically. Association does not establish causation.
Historical mean and volatility, IID normal log returns, fixed seed. This model does not capture market regimes or extreme tails; its bands are not forecast confidence intervals.
Cash flow over time
Compounding
Bond price under yield shifts
IMPLIED FORWARD RATE
—
Derived from zero-coupon spot rates, with annual compounding. It is a no-arbitrage implied rate, not a prediction of the next central-bank decision.
Observations can lag markets and be revised. Policy schedules open at the source; future rate decisions are not known in advance.
Cash flows & present value
NPV at different hurdle rates
| Year | Net cash flow | Present value | Cumulative PV |
|---|
Return measures & sensitivity
Cash-flow improvement × hurdle rate
An improvement adds the selected percentage of each operating flow's absolute value; a negative flow becomes less negative. Initial investment, salvage and working capital stay fixed.
Nominal value Today's purchasing power
Required payment
| Scenario | Local real return | Real USD return | Nominal end value | Purchasing power | Gap to real target |
|---|
Target & break-even
Annual values
| Year | Nominal USD | Today’s USD | Real target |
|---|
Stress test · return × US inflation
Selected scenario at ±1 and ±2 percentage points. FX stays fixed. Cells show ending purchasing power.
| Economy | Real GDP growth | CPI inflation | Nominal lending rate | Real lending rate |
|---|
Annual historical context, not a project-specific cost of capital. World Bank real lending rates use the GDP deflator; your modeled real hurdle uses expected inflation. Lending terms differ across countries. Missing observations remain blank; latest mode can mix years, shown beside every value.
Price under yield changes
Promised cash flows
| Year | Coupon | Principal | Present value |
|---|
Five-year operating model
| Year | Revenue | EBITDA | Cash tax | FCFF |
|---|
| Case | Growth | Margin | Enterprise value | Equity / share |
|---|
Cash-flow calculation
| Year | EBIT | − Tax | + D&A | − Capex | − ΔNWC | = FCFF |
|---|
Priority & pari passu recovery
| Debt instrument | Rank | Claim | Payment | Recovery |
|---|
Rank 1 is paid first. Instruments at the same rank share the available value pro rata.
Maturity wall
Daily return distribution
Distribution detail
| Series | Daily mean | Daily SD | Annual SD | Skewness | Excess kurtosis |
|---|
SD is sample standard deviation (n−1); annual SD = daily SD × √252. Skewness and excess kurtosis use finite-sample bias corrections and are not annualized. Normal-reference excess kurtosis is 0. Undefined moments display —. These are observed returns, not the shrinkage covariance estimate.
Research paths across the major asset classes. Coverage labels distinguish daily price lookup, external sources and assumption-led models.
Search, then select the correct exchange and instrument.
Choose an instrument above to load its history.
Capital structure & recovery
| Priority | Claims · USD | Paid · USD | Recovery | Shortfall |
|---|
FOUR EDITABLE STARTERS
Excel model templates
Company valuation, debt recovery, project appraisal and portfolio risk. Open the workbook in Excel, change the blue inputs, and follow each calculation.
Company DCF
Forecast cash flow, discount it, add terminal value, and bridge enterprise to equity value.
Debt recovery
Order the claims. Allocate recovery. Compare your entry price with discounted proceeds.
Project appraisal
Enter investment and yearly cash flows. Read NPV, IRR, and the nominal or real hurdle.
Portfolio risk
Replace aligned return observations. Calculate covariance, volatility and historical Sharpe.
| Action | Keys | Area |
|---|
Desktop Excel, US keyboard. Your Mac may need Fn with function keys. Formula-editing shortcuts depend on context.
Model-building sequence
- State the decision. Choose one output: a value, a return or a funding need.
- Set the units and dates. Keep currencies, annual rates and cash-flow periods consistent.
- Enter sourced inputs. Record the source and observation date beside each assumption.
- Build the calculation. Reference cells; avoid burying assumptions inside formulas.
- Test a known case. Try a zero rate, full recovery or equal-weight portfolio.
- Change one driver. Check direction and magnitude, then save a dated version.
Useful formulas
Scratch-sheet examples: rate in B1, year-0 flow in B3, years 1–5 in B4:B8 and dates in column A. For statistics, use returns in B3:C14. These are separate examples, not model-kit cell addresses.
=NPV(B1,B4:B8)+B3Year 0 is outside NPV.
=IRR(B3:B8)Equal periods. Multiple sign changes can create multiple IRRs.
=XNPV(B1,B3:B8,A3:A8)Use actual dates for irregular cash flows; pair with XIRR.
=STDEV.S(B3:B14)Sample standard deviation; align observations first.
=COVARIANCE.S(B3:B14,C3:C14)Use the same dates and return frequency.
=KURT(B3:B14)Sample excess kurtosis; normal reference is zero.
Workbook inputs are fictional teaching examples. No live feed, macros or account connection. Changes in Excel do not sync back to your portfolios.
Choose a rule and run it. The later portion of the history is reported separately as a holdout.
Attribution
Exposure over the full run
| Segment | Days | Total return | CAPM alpha / yr | Sharpe | Max. drawdown |
|---|
Interpretation & data limits
Rules run long / cash. No leverage or short sales. See Sources for timing, costs and statistical assumptions.
Review the execution
Freeze this study and replay its allocations, costs and risk halt.
Build the legs and calculate the payoff.
Position payoff
Position Greeks
Local model sensitivities at the current inputs. Greeks change as price and time change.
| Leg | Contracts | Strike | Entered premium | European value | Delta / share |
|---|
Bring a tested rule here
Run a sandbox study, then choose Open paper replay.
Paper equity
Current allocation
| Asset | Value | Weight |
|---|
| Execution close | Signal close / rule | Paper notional | Costs |
|---|
Runs only while this tab is open and visible. Navigation pauses it; saved replays reopen paused. This uses a frozen historical study, with adjusted-price notionals rather than broker share fills.
Options, Futures, and Other Derivatives
John C. Hull. Connect no-arbitrage pricing, volatility, option spreads and delta hedging to explicit payoff and sensitivity calculations.
Publisher reference ↗Efficiently Inefficient
Lasse Heje Pedersen. A useful research question: what economic friction pays for the strategy’s information, financing and trading costs?
Book & author reference ↗Time series momentum
Moskowitz, Ooi & Pedersen (2012) study trend across futures and forwards. Our moving-average equity recipe is a teaching variation; it does not reproduce their portfolio or evidence.
Read the research ↗Quality minus junk
Asness, Frazzini & Pedersen. Test profitability, growth, safety and payout with point-in-time fundamentals. A price-only screen cannot reproduce this factor.
Methodology & dataset ↗Backtest overfitting
Bailey, Borwein, López de Prado & Zhu. Selecting a winner from many trials can manufacture an impressive backtest. Keep a trial log and reserve new data for confirmation.
Read the paper ↗The sandbox shows a chronological holdout and HAC uncertainty. It does not calculate the paper’s probability-of-overfitting statistic.
The fund prospectus
Tulane’s FINE 7350 uses academic research, applied cases and a hypothetical hedge-fund prospectus. Use the memo to state mandate, edge, implementation, risk, capacity and failure conditions.
Tulane course catalog ↗Based on the public course description and your previously mapped fund/quality concepts. Private hedge-fund class notes have not been located.
| Family | Question to test | Risks and additional data | Workspace |
|---|---|---|---|
| Equity long / short | Is performance stock selection or common factor exposure? | Point-in-time fundamentals, delistings, borrow availability and fees. | |
| Systematic trend | Does a lagged rule survive costs and different periods? | Whipsaw, regime change and contract rolls. Sandbox uses equities, long / cash. | |
| Relative value | Is a spread economically linked and stable? | Correlation alone does not imply cointegration; hedge drift, liquidity and financing. | |
| Event driven / distressed | What catalyst changes the value available to each claim? | Timing, failed catalysts, collateral, covenants, priority and pari passu treatment. | |
| Global macro | Which rate, inflation and currency scenario is mispriced? | Country risk, duration, policy surprises, FX and crowded positioning. | |
| Volatility / options | Does the premium compensate for convexity and tail exposure? | Volatility surface, skew, assignment, margin, gaps and transaction costs. |
Saved portfolios & transactions
Personal and team portfolios record paper purchases and sales. Holdings use weighted-average cost; realized gains are sale proceeds less the released cost basis. Reference prices are manually entered with their observation dates; no reference price means value and unrealized profit are unavailable. Execution timestamps are supplied by the contributor; save timestamps are recorded by the server. Transactions cannot be overwritten. Fees, dividends, splits and currency conversion are not modeled. All amounts are USD.
Relationships
Pearson correlation and rolling correlation use daily adjusted returns. The pair regression is Y = intercept + beta × X, with HC3 robust standard errors. A positive lag pairs earlier X returns with later Y returns. The full matrix always uses same-day returns. Association does not establish causation or forecast performance.
Historical portfolio calculations
Daily adjusted prices aligned on common dates. No forward filling. Arithmetic annual return and covariance use 252 trading days. Performance is a constant-weight, daily-rebalanced hypothetical portfolio with no fees, tax or slippage. This historical model is separate from your saved transactions on the Portfolio tab.
Risk & optimization
Sample or Ledoit–Wolf covariance; long-only SLSQP optimization with your position cap. Historical Sharpe uses daily excess returns. Model Sharpe uses estimated annual return minus your annual risk-free rate. Maximum drawdown includes the initial value; VaR and CVaR are historical daily losses at 95%.
Regression & simulation
OLS on excess returns with HC3 robust standard errors. The simulation samples IID normal log returns fitted to the same history. Optimized results are in-sample estimates, with no claim of future performance.
Corporate finance
NPV discounts year-end after-tax unlevered flows plus final salvage and working-capital recovery; year 0 is initial investment plus working capital. All future amounts use the selected cash-flow basis. Nominal and real discount rates are matched using (1 + nominal) / (1 + inflation) − 1. IRR lists all numerically verified real roots above −100%; nonconventional projects can have multiple roots. MIRR uses the matched hurdle for both financing and reinvestment. Fractional payback assumes uniform receipts within each year; discounted payback interpolates annual present values. Profitability index is shown only for conventional cash flows.
NYU Stern · consistent DCF inputs ↗
Time value & distributions
TVM uses nominal APR divided by periods per year, with fixed payments at the start or end of each period. Purchasing-power values use annual expected inflation. Historical distribution statistics use sample variance (n−1), bias-corrected Fisher–Pearson skewness and Fisher excess kurtosis. No annualization of higher moments.
SciPy · excess kurtosis ↗ · Skewness ↗
Global comparisons
World Bank annual historical observations retain their year, retrieval timestamp and dataset update date. Six-hour source cache; missing and forecast observations are not substituted. CPI inflation and GDP-deflator-adjusted real lending rates are different measures. Country lending terms differ; these are context, not a risk-matched corporate hurdle rate.
Valuation & rates
DCF uses unlevered free cash flow, WACC and a Gordon terminal value. Rates use annual compounding; bonds use semiannual coupons. A forward rate is implied by spot rates, not a Fed or ECB prediction.
Markets & distressed debt
Instrument search and history use Yahoo via yfinance, cached up to 15 minutes. Prices are adjusted daily bars in provider quote units; the current session may be incomplete. No order-book, options chain, individual-bond or private-asset feed is connected. The debt model distributes value after cash and costs down an assumed priority order, capped at claims; your selected-class holding receives its pro-rata share. Recovery NPV discounts one final cash payment at your hurdle. Annualized recovery return is (proceeds / cost)^(1 / years) − 1. This is a scenario, not a legal determination or probability-weighted forecast.
Corporate bonds
Fixed coupons, bullet redemption and settlement on a coupon date, so accrued interest is zero. Solve price = discounted promised coupons + principal for nominal annual YTM with the chosen frequency. Macaulay duration is PV-weighted payment time; modified duration divides it by 1 + YTM / frequency. DV01 = price × modified duration × 0.0001. Convexity is the normalized second price derivative. Yield shocks reprice all payments. YTM is not an expected default-adjusted return; the simple benchmark difference is not OAS or a Z-spread.
FINRA · bond yield and return ↗
Capital structure & pari passu
Rank 1 is first in an assumed single recovery estate. Claims with the same rank are pari passu in this model and share that rank's available recovery pro rata by face amount, capped at face. A common seniority label does not establish equal recovery across issuers, collateral pools or guarantors. Structural and effective subordination, intercreditor terms, accrued claims and priority expenses need separate analysis. Residual is after modeled debt, before omitted claims.
Market debt value = face × price / 100. CAPM cost of equity = risk-free rate + levered beta × equity risk premium. WACC uses market-value debt, common equity and preferred weights; debt cost is your forward rate, tax-adjusted, not its coupon. Preferred capital affects WACC only here. Leverage uses face debt / positive EBITDA. Coverage uses EBITDA / annual coupons; it is not a liquidity forecast. Maturities are principal due by relative year, assuming no refinance.
Example prospectus · ranking and subordination ↗
Company scenarios
Forecast five years of revenue, EBITDA, D&A, positive-EBIT cash taxes, capex and incremental net working capital. FCFF = EBIT − taxes + D&A − capex − ΔNWC. Terminal value uses sustainable year-5 FCFF, perpetual growth and WACC. Equity bridge adds cash and deducts debt value and other non-equity claims; a negative bridge remains visible while common-equity value is floored at zero. Downside and upside shift growth and margin jointly, with no assigned probability. Transferring WACC or EV copies an assumption for review; it does not silently change other models.
Course research lens
The supplied Chapter 13 examples inform the cost-of-debt, CAPM, market-weight WACC and risk-matched project tools. The distress lecture motivates separating weak operations, near-term liquidity, unexpected liabilities and confidence shocks. A low price or rating alone does not explain recovery.
The Quality Minus Junk material motivates checking profitability, growth, safety and payout. Compare margins, returns on capital, cash generation and leverage before drawing a conclusion. This edition does not claim to reproduce AQR's cross-sectional factor or compute a licensed quality score.
The analyst material motivates dated targets, explicit assumptions, revisions and comparable EPS definitions. EPS surprise is actual minus forecast EPS on the same accounting basis; it is different from a standardized unexpected-earnings statistic. Recommendations and targets are hypotheses to test against later outcomes, not facts about future prices.
The fund syllabus connects quantitative screening, qualitative issuer research, relative valuation, factor exposure and portfolio risk. The original classroom decks, exams, answer sheets and syllabus are not distributed by this application.
AQR · Quality Minus Junk methodology ↗
Excel models
The downloadable kit contains four independent, editable teaching models. Blue inputs are fictional; black formulas calculate within a sheet. DCF uses annual unlevered cash flows and WACC. The project template uses five end-of-year flows plus initial investment. The portfolio template uses aligned monthly decimal returns and 12 periods per year. It is not linked to saved transactions. Sources and limits accompany each model.
Global scenario model
Inputs are annual effective rates, held constant for the selected horizon. Local real return = (1 + nominal return) / (1 + local inflation) − 1. USD nominal return = (1 + nominal return) × (1 + annual currency change) − 1. Real USD return divides that gross USD return by (1 + US inflation), then subtracts 1. Positive currency change means that one unit of the scenario currency buys more USD. The model normalizes the initial exchange conversion to the entered USD capital; it does not fetch exchange rates.
Future USD value = starting USD × (1 + USD nominal return)^years. Purchasing power divides future USD by (1 + US inflation)^years. The target compounds starting capital at the entered real USD target. Required nominal return = (1 + real target) × (1 + US inflation) / (1 + FX change) − 1. Break-even FX = (1 + US inflation) / (1 + nominal return) − 1. Sensitivity changes the scenario return and US inflation by ±1/±2 percentage points, holding FX fixed. No contributions, tax, fees, credit losses, volatility or hedging costs are modeled. Required rates can exceed the input range; they are arithmetic thresholds, not available products.
Initial cases are illustrative, not estimates or policy forecasts. Published World Bank observations are a separate historical reference and never overwrite the scenarios. The World Bank real lending series uses the GDP deflator, not CPI; its nominal lending rate is a borrowing-cost statistic, not an investment return.
Federal Reserve · exact real-rate calculation ↗ · World Bank · real lending definition ↗
Alpha sandbox
Daily adjusted closes, USD assumed, up to eight assets and 1,250 common observations. Signals use information through the prior close; allocation changes execute at the next close and earn returns thereafter. Prices are treated as adjusted notional units, not executable share quotes. Existing positions drift between rebalances. Cash compounds at (1 + annual rate)^(1/252) − 1. Moving-average trend holds eligible assets equally; momentum ranks positive lookback returns; reversion ranks negative lookback returns. Ranked recipes hold up to N names. Target exposure and per-asset caps apply at rebalancing; unused weight remains in cash. Drift can exceed the target cap between rebalances. No shorting or leverage.
Commission plus slippage is charged on both buy and sell notionals. A self-financing equation deducts costs before sizing the new allocation. A close-based drawdown halt liquidates after the loss has occurred and persists to the end; gaps and fees can exceed its threshold. No intraday fills, market impact, taxes, delistings or borrowing constraints are simulated. The benchmark is buy-and-hold, without costs. End values are marked, not liquidated.
Warm-up is excluded. The research/holdout split is chronological with at least 63 returns per segment; portfolio and risk state carry across the split. No parameter optimizer is used. CAGR compounds net returns with 252 sessions/year. Sharpe uses mean daily excess return divided by sample daily deviation, times √252. CAPM regresses net strategy excess returns on benchmark excess returns with an intercept; alpha is the daily intercept × 252. HAC covariance uses five lags with t-based 95% intervals. This is single-factor attribution, not a Fama–French or QMJ model. Repeated trials and chosen surviving tickers invalidate strong discovery claims. Holdout results are visible and therefore no longer untouched once inspected. No multiple-testing adjustment is computed.
Backtest overfitting ↗ · Trend research ↗
Options workbench
Black–Scholes–Merton European options with continuous annual rates and dividend yield, one expiry, 365-day years, constant volatility. Entered premiums and model values are distinct. Positive contracts are long; negative contracts are short. Expiry P/L adds stock gains and each signed intrinsic value, subtracts signed entry premiums and per-contract fees. Maxima, minima and break-even points use strike breakpoints and the terminal slope over nonnegative underlying prices. Unlimited tails are reported explicitly. Current model P/L holds remaining time constant across the price curve. Neither curve accrues premium financing, stock dividends or exercise/exit costs.
Greeks aggregate signed contracts × multiplier; stock contributes to delta only. Gamma is delta change per $1, vega and rho are dollars per one percentage point, theta is dollars per calendar day. American exercise, assignment, volatility skew, live chains, liquidity and margin are not modeled. Stock and option contract sizes must match for a covered position. Presets use illustrative $100 underlying prices and manually entered premiums.
Hull textbook ↗ · OIC bull call spread ↗
Paper automation
The engine replays the frozen sandbox path locally in the browser. Start, pause, single-step and run-to-end controls reveal only the current replay position. It pauses on navigation or when the tab becomes hidden, and saved snapshots reopen paused. It is historical replay, not a live price listener, server scheduler or broker executor. It does not write to your personal/team transaction books. Notebook saves remain personal under the existing access rules.
Market data
Sample mode is synthetic. Yahoo is adjusted daily history, not a real-time feed. Uploaded CSV adjustments and currency are supplied by you. FRED observations retain their individual dates and source links. Hosted distribution needs appropriate provider permissions and a market-data plan.