OLIVER SCOTT
Home / Research agenda

Research agenda

Twelve questions I want to answer, each written as a short proposal: the hypothesis, the data, the method and what could fool me. Most use public Australian data, and several extend the mining study.

R-01

Placement discounts and what follows

Scoping
Question
How do ASX small caps trade after discounted share placements, and does a short cash runway predict a deeper discount?
Hypothesis
Discounts are larger for companies with fewer quarters of funding (Appendix 5B item 8.7), and returns after the placement drift negative for low-runway issuers.
Data
ASX announcements (placement, Appendix 2A/3B), quarterly Appendix 5B cash and runway, daily prices.
Method
  • Parse discount to last close and to 5-day VWAP from each placement announcement.
  • Regress discount on runway, size, liquidity and recent run-up.
  • Event study of returns to 60 days, split by runway terciles.
Pitfalls
  • Placements announced after trading halts: day-0 price is stale.
  • Survivorship: issuers that fail later drop out of free data.
  • Attached options and tranche approvals change the true discount.
Why it matters
Connects directly to the cash-runway filter in my mining study and tests its rationale.
Starting point
Wruck (1989), Journal of Financial Economics: equity ownership concentration and private sales.
R-02

Director buying in small caps

Planned
Question
Do on-market share purchases by directors (Appendix 3Y) predict returns in ASX small caps?
Hypothesis
Open-market purchases predict positive abnormal returns over 3 to 6 months, concentrated in small, thinly covered firms; sales carry little information.
Data
Appendix 3Y notices (date, type, value), prices, market capitalisation.
Method
  • Classify notices: on-market buy, sell, plan or option exercise.
  • Calendar-time portfolios of buys by size and value bought.
  • Control for momentum and size; check overlap with results season.
Pitfalls
  • Lodgement delay: trade date and notice date differ.
  • Exercises and dividend-plan shares are not signals.
  • Thin trading inflates measured returns.
Why it matters
A clean, public information signal that large funds largely ignore in micro caps.
Starting point
Lakonishok and Lee (2001), Review of Financial Studies: are insider trades informative?
R-03

Short interest as a signal

Planned
Question
Do rising short positions in ASIC's daily short position reports predict underperformance once borrow costs and liquidity are accounted for?
Hypothesis
Stocks with the largest increases in reported short positions underperform over the following month, but the effect weakens after realistic borrow costs.
Data
ASIC short position reports (published with a short lag), prices, liquidity.
Method
  • Build short-interest level and change, aligned to publication date to avoid look-ahead.
  • Decile sorts and Fama-MacBeth regressions with size, momentum and value controls.
  • Net of estimated borrow fees for hard-to-borrow names.
Pitfalls
  • Publication lag versus position date.
  • Index arbitrage and hedging inflate shorts without negative views.
  • Borrow costs are not public; need assumptions and ranges.
Why it matters
Tests whether a widely available signal survives frictions, a recurring theme in practical quant research.
Starting point
Boehmer, Jones and Zhang (2008), Journal of Finance: which shorts are informed?
R-04

Franking credits and the ex-dividend drop

Planned
Question
On the ASX, how much of a franking credit does the share price drop by on the ex-dividend date, and who sets the price?
Hypothesis
Prices drop by the cash dividend plus a fraction of the franking credit well below one, and the fraction rises for stocks held mainly by domestic investors.
Data
Dividend amounts and franking percentages, ex-dates, opening and closing prices.
Method
  • Drop-off ratio regressions of the price fall on cash and franking components.
  • Split by ownership proxies, dividend yield and period (tax-rule changes).
  • Check the pre-ex-date run-up studied in my event framework.
Pitfalls
  • Market moves on the ex-date dominate small dividends; need market adjustment.
  • Bid-ask bounce around the open.
  • Special dividends behave differently.
Why it matters
A uniquely Australian question with a precise, testable prediction.
Starting point
Walker and Partington (1999), Accounting and Finance: the value of dividends from cum-dividend trading.
R-05

Closing auction imbalances

Planned
Question
Do order imbalances in the ASX closing single-price auction predict next-day returns?
Hypothesis
Large closing buy imbalances, especially on index rebalance days, are partly reversed the next morning.
Data
Closing auction prices and volumes; indicative imbalance where available; index event calendars.
Method
  • Measure closing auction price versus last continuous price and auction volume share.
  • Regress overnight and next-day returns on the auction premium.
  • Separate index rebalance days from ordinary days.
Pitfalls
  • Imbalance data may need a paid feed.
  • Overnight news confounds next-day returns.
  • Transaction costs can exceed small reversals.
Why it matters
Market microstructure with direct links to execution, relevant to trading and dev roles.
Starting point
Market-structure literature on closing auctions and liquidity provision.
R-06

Rio Tinto plc versus Rio Tinto Ltd

Scoping
Question
How far and how long do the London and Sydney lines of Rio Tinto's dual-listed company structure deviate from parity, and what drives it?
Hypothesis
Deviations are mean-reverting with a half-life of days to weeks and widen with dividend tax differences, FX moves and index flows.
Data
Daily prices of Rio Tinto plc (LSE) and Rio Tinto Ltd (ASX), GBP/AUD, dividends.
Method
  • Build the price ratio in a common currency; test for stationarity.
  • Estimate an Ornstein-Uhlenbeck half-life and how it varies over time.
  • Explain deviations with franking, FX, and index-event dummies; check BHP's history before its 2022 unification.
Pitfalls
  • Non-overlapping trading hours: use synchronised prices or overlapping windows.
  • Franking credits make exact parity the wrong benchmark.
  • Shorting costs on either line.
Why it matters
A classic 'Siamese twins' relative-value problem that bridges the ASX and London.
Starting point
Froot and Dabora (1999), Journal of Financial Economics: how are stock prices affected by the location of trade?
R-07

Gold miners as leveraged gold

Planned
Question
How large and how stable is ASX gold miners' sensitivity to the gold price, and do hedging programmes explain the differences?
Hypothesis
Betas to gold are above one on average, vary with operating leverage, and fall for miners with large hedge books.
Data
Daily gold price in AUD and USD, miner prices, hedge disclosures from quarterly reports.
Method
  • Rolling regressions of miner returns on gold and market returns.
  • Cross-sectional regression of beta on costs (AISC), hedging and size.
  • Out-of-sample: does estimated beta predict next quarter's beta?
Pitfalls
  • AUD/USD moves mix with gold in AUD terms.
  • Hedge disclosures are irregular.
  • Estimation noise in short windows.
Why it matters
Informs how the commodity-shock signal in my study should size positions.
Starting point
Tufano (1998), Journal of Finance: stock price exposure and financial engineering in gold mining.
R-08

Index rebalance effects, then and now

Planned
Question
Do additions to and deletions from the S&P/ASX 200 still move prices, and has the effect shrunk as indexing grew?
Hypothesis
Announcement and effective-date returns have fallen over time, and much of the move now happens before the announcement.
Data
S&P Dow Jones Indices announcement history, prices, volumes, index fund ownership trends.
Method
  • Event study around announcement and effective dates.
  • Subperiods to track decay.
  • Reversal after the effective date as a liquidity-provision return.
Pitfalls
  • Pre-announcement drift from predictable changes.
  • Small samples per quarter.
  • Confounding results releases.
Why it matters
Tests a well-known anomaly for decay, the honest question every live strategy faces.
Starting point
Shleifer (1986), Journal of Finance: do demand curves for stocks slope down?
R-09

ASX ETF premiums and discounts

Planned
Question
Do premiums and discounts to net asset value on ASX-listed ETFs mean-revert, and is any of it tradable after spreads?
Hypothesis
International ETFs show larger, more persistent deviations because underlying markets are closed during ASX hours, but spreads absorb most of the profit.
Data
Daily ETF prices and published NAVs, underlying index futures, spreads.
Method
  • Measure premium relative to a time-aligned fair value, not stale NAV.
  • Estimate mean reversion by ETF type.
  • Simulate a threshold strategy with realistic spreads.
Pitfalls
  • Stale NAV creates fake premiums.
  • Market makers already arbitrage large deviations.
  • Survivorship among closed ETFs.
Why it matters
Shows the difference between apparent and real mispricing, a core skill in trading roles.
Starting point
Petajisto (2017), Financial Analysts Journal: inefficiencies in the pricing of exchange-traded funds.
R-10

Volatility risk premium on the ASX 200

Planned
Question
How large is the gap between implied and realised volatility on ASX 200 options, and what does selling it with defined risk earn after costs?
Hypothesis
Implied exceeds realised on average, but the premium is concentrated in calm periods and a defined-risk strategy gives up much of it.
Data
ASX 200 implied volatility (index level or option quotes) and realised volatility from daily or intraday prices.
Method
  • Compare 30-day implied with the following 30-day realised volatility.
  • Simulate put spreads and iron condors with realistic spreads and margin.
  • Analyse tail months and drawdowns.
Pitfalls
  • Option quote data may need a paid source.
  • Overlapping windows inflate t-statistics.
  • Rare crashes dominate results.
Why it matters
Brings my options pricing project to real data and links to the systematic option-selling idea.
Starting point
Carr and Wu (2009), Review of Financial Studies: variance risk premiums.
R-11

AUD carry and commodity links

Planned
Question
How much of the Australian dollar's behaviour is explained by carry, momentum and the prices of iron ore and gold?
Hypothesis
Commodity returns explain a meaningful share of AUD moves, and carry returns crash in risk-off episodes.
Data
AUD crosses, interest rate differentials, iron ore and gold prices.
Method
  • Factor regressions of AUD returns on commodities, carry and global risk proxies.
  • Time-varying exposure with rolling windows.
  • Out-of-sample forecasting test against a random walk.
Pitfalls
  • Random-walk benchmark is hard to beat in FX.
  • Iron ore price series have changed contracts over time.
  • Look-ahead in rate data revisions.
Why it matters
Macro breadth, and a check on the FX component hidden inside every AUD-priced commodity signal.
Starting point
Lustig, Roussanov and Verdelhan (2011), Review of Financial Studies: common risk factors in currency markets.
R-12

Intraday patterns on the ASX

Planned
Question
Are there repeatable patterns in returns and volume around the ASX open, midday and close, and do they persist at the same half-hour across days?
Hypothesis
Returns in a given half-hour predict returns in the same half-hour on following days, and volume is U-shaped with a large closing auction share.
Data
Intraday bars (collected by my live system), auction volumes.
Method
  • Build half-hour returns; test same-interval continuation across days.
  • Separate auction prints from continuous trading.
  • Cost-adjust any implied strategy.
Pitfalls
  • Short history from self-collected data.
  • Staggered open by ticker group distorts the first interval.
  • Microstructure noise in small caps.
Why it matters
Uses data my own system is already recording, so it grows with the forward test.
Starting point
Heston, Korajczyk and Sadka (2010), Journal of Finance: intraday patterns in the cross-section of returns.

Status moves from Planned to Scoping once I've checked the data, then to a full write-up like the mining research note.